Only a small number of people in Karnataka will ever earn a living competing. Far more can earn one building. This page is a map — how games are actually made, what each discipline does, what AI is genuinely changing, and how someone gets from "I play games" to "I ship games". The industry content here is general and verifiable; the academy described further down is a plan, and is marked as one.
Karnataka hosts roughly 300 AVGC studios, twelve gaming-focused institutes, five animation institutes and three VFX institutes — a meaningful share of India's specialised training capacity in this field. The state's AVGC-XR Policy 3.0 (2024–2029) commits around ₹150 crore, a Gaming Centre of Excellence, an AVGC-XR park and a dedicated KITVEN venture fund, with a jobs target in the tens of thousands and an ambition to earn most sector revenue from exports.
Nationally the demand side is equally clear: India's gamer base crossed roughly 520 million players in 2025, with over 100 million people playing daily. Bengaluru GAFX — the state's annual AVGC-XR summit — drew over 200 speakers across 100+ sessions in its 2026 edition, themed around the convergence of human creativity with AI and real-time engines.
What is still thin is the middle: a structured, free, district-accessible route from a curious sixteen-year-old in Haveri to a junior engine programmer in a Bengaluru studio. Building that route is what the academy described further down is meant to do — it does not exist yet.
Most people picture game development as "coding a game". In practice it is a production pipeline with distinct phases, gates and failure modes. Understanding the shape of it is the first thing that separates a hobbyist from a professional.
What is the game, who is it for, why would they keep playing? Output is a one-page pitch and a core-loop statement — not a design document.
Ugly, fast, disposable. Proves the core loop is fun with placeholder art. Most game ideas die here, and that is the prototype doing its job.
Build the vertical slice: one section of the game at final quality. This is where the real technical and art pipelines get established.
The long middle. Content volume gets built against the pipeline established in pre-production. Discipline here beats brilliance.
Beta to gold. Feature-locked, bug-burning, performance-tuning, and platform compliance — the unglamorous phase that decides reviews.
For any modern competitive or mobile title, shipping is the start. LiveOps is a permanent discipline with its own team, cadence and economics.
"Game developer" is not a job. These are. Most people are strong in one, competent in two adjacent ones, and literate in the rest.
Rendering, physics, memory, threading, networking, tools. Deep C++, low-level performance work. The hardest to learn and the hardest to replace.
Character controllers, combat, abilities, AI behaviour, state machines. Sits closest to design and iterates fastest.
Client-server architecture, replication, rollback and lag compensation, matchmaking. Essential for competitive titles and chronically under-supplied in India.
Progression curves, reward schedules, matchmaking ratings, in-game economies. Mathematics and psychology in equal parts.
Space, pacing, sightlines, encounter composition. In competitive shooters this is the single largest determinant of whether a map is played or skipped.
The bridge between art and engineering: shaders, rigs, tooling, optimisation, procedural asset generation. Among the most employable skills in the market.
Concept, 3D modelling, texturing, rigging, animation, VFX, lighting. Karnataka's deepest existing talent pool, thanks to its animation and VFX industry.
SFX, adaptive music, mixing, implementation in Wwise or FMOD. Massively undervalued: audio is half of perceived game feel.
HUD, menus, readability under pressure, input mapping, accessibility. In an esports title the HUD is a competitive instrument, not decoration.
Test planning, reproduction, automation, performance profiling, compliance. The standard entry point into the industry, and a genuine career in its own right.
Version control at scale, build farms, CI/CD, platform deployment, telemetry infrastructure. Invisible when it works.
Scheduling, scope, risk, budgets, milestone delivery, store relationships, marketing beats, community. The reason projects finish.
Five ideas that carry most of the weight in practical game design. Learn these before learning any engine.
Every game reduces to a repeating cycle of action → feedback → reward → upgraded ability → harder action. If that loop isn't satisfying in thirty seconds with placeholder art, no amount of content will save it.
Example — battle royale: drop → loot → engage → survive/upgrade → shrinking circle forces the next engagement. The circle is the design. Without it the loop stalls into passive camping.
The MDA framework: Mechanics are the rules you write; Dynamics are the behaviour that emerges when players meet those rules; Aesthetics are the feeling that results. Designers author mechanics but are judged on aesthetics — and the gap between them is where design actually happens.
Practical consequence: you cannot design a feeling directly. You can only build mechanics, observe the dynamics, and iterate.
A competitive game needs a low floor (a new player can participate on day one) and a high ceiling (mastery keeps revealing depth for years). Games that fail commercially usually have a high floor; games that fail competitively usually have a low ceiling.
Balance is not "everything equally strong". It is "every option strong in some identifiable situation, and no option strong in all of them".
Also called "juice": input latency, animation weight, camera shake, hit-stop, particle response, audio layering, controller rumble. Individually trivial, collectively the difference between a game that feels dead and one that feels alive. Budget real time for it — it is not polish, it is a feature.
Watch people play without helping them. Where they hesitate is a design problem, not a player problem. Five silent observations will teach you more than fifty survey responses.
The engine question gets far more attention than it deserves. Pick the one whose default strengths match your project and whose community you can reach in your language.
| Engine | Language | Best suited to | Learning curve | Licensing model | Notes for Karnataka teams |
|---|---|---|---|---|---|
| Unity | C# | Mobile, 2D, mid-size 3D, AR/VR | Moderate | Tiered, revenue-linked | Deepest local hiring pool. The default for Indian mobile studios and the right first engine for most students. |
| Unreal Engine 5 | C++ / Blueprints | High-fidelity 3D, shooters, cinematics, virtual production | Steep | Royalty above a revenue threshold | Strong overlap with Karnataka's VFX and animation industry. Blueprints let designers prototype without code. |
| Godot | GDScript / C# | 2D, small-to-mid 3D, tooling, education | Gentle | Free and open source (MIT) | Best classroom engine: no licence cost, tiny download, runs on low-spec machines common in district labs. |
| Defold | Lua | Lightweight mobile and web 2D | Gentle | Free | Very small build sizes — meaningful when your audience is on entry-level Android and constrained data. |
| Roblox Studio | Luau | Social/UGC games, first-time creators | Very gentle | Platform revenue share | Excellent absolute first step for school chapters. Publishing to a real audience in week one is a powerful motivator. |
| Custom / in-house | C++ / Rust | Specialised performance or platform needs | Very steep | N/A | Almost never the right choice for a first project. Learn one commercial engine deeply first. |
Separating the genuine shifts from the marketing. AI has been in games since the 1970s as behaviour trees and pathfinding; what changed recently is generative and learned systems entering the production pipeline.
Concept ideation, texture variation, LOD generation, retopology assistance, animation retargeting and cleanup, lip-sync, upscaling. These reduce hours per asset, which changes what a ten-person studio can attempt.
Effect: smaller teams, wider scope. Not fewer artists — different artists.
Machine translation with human review makes it economic to ship in Kannada, Tamil, Telugu, Hindi, Marathi and Bengali simultaneously. For an Indian studio this is not a nicety — it is the difference between a metro audience and a national one.
Effect: Indian-language games become commercially viable earlier.
Reinforcement-learning agents that play thousands of matches overnight to find broken strategies, unreachable geometry, degenerate economies and balance outliers before humans ever see the build.
Effect: faster balance iteration; QA moves up the value chain.
Agents trained rather than scripted. Promising for driving lines, combat opponents and training bots — but harder to debug, harder to certify, and prone to behaviour designers did not intend. Adoption is cautious for good reasons.
Effect: better practice bots; slower adoption in shipped campaigns.
NPCs that respond conversationally rather than from a script tree. The technology works; the hard problems are tone control, canon consistency, safety, latency and cost per interaction on mobile hardware.
Effect: real potential, currently more demo than shipped feature.
PCG is decades old; what is new is learned models guiding it — generating level layouts, quest structures or economies that satisfy designer-authored constraints rather than pure randomness.
Effect: designers curate generators instead of hand-placing content.
Behavioural models that flag aim patterns, reaction-time distributions and input signatures inconsistent with human play. Directly relevant to KEF's integrity code — though every flag remains a human decision, never an automatic sanction.
Effect: detection improves; due process must not be automated away.
Automated VOD review, positioning heatmaps, economy decision grading, opponent tendency modelling. This is the layer that turns raw match data into coachable feedback — and it is buildable by a student with Python and a data feed.
Effect: coaching becomes evidence-based, and a new job category appears.
Automated observer camera selection, highlight detection, multi-language commentary assistance, clip generation. Directly enables KEF's Kannada-first broadcast goal at district-event budgets.
Effect: small events get broadcast production they could never staff.
No cohort has run and no enrolment is open. This is the curriculum design, published so that colleges, studios and teachers can tell us what is wrong with it before it is delivered. The design principle: every track ends with a shipped artefact — a playable build and a public demo — and no certificate is issued without one.
| Week | Topic | What you build | Concept introduced |
|---|---|---|---|
| 01 | What is a game? | Paper prototype of an existing mobile game | Core loop, pillars, MDA |
| 02 | Engine orientation | Moving square with collision | Scenes, nodes, the game loop, delta time |
| 03 | Input & feel | Platformer controller with coyote time | Game feel, input buffering, juice |
| 04 | Systems & state | Score, lives, win/lose states, menus | State machines, UI flow |
| 05 | Content & iteration | Three levels, playtested with five strangers | Grey-boxing, silent observation, iteration |
| 06 | Ship it | Public build with a title screen and credits | Build pipeline, scope cutting, shipping |
Designed to be delivered in Kannada and English, and to run on machines with 8 GB of RAM and integrated graphics — deliberately, so that a district college lab can host it without buying new hardware. If that constraint is unrealistic for your institution, we want to know.
Roles, typical entry routes and indicative compensation for the Indian market. Ranges are broad and vary enormously by studio type, city and whether the employer is a domestic mobile studio, a global AAA outsourcing arm or a product company.
| Role | Entry route | Core skills | Indicative entry (₹ LPA) | Indicative senior (₹ LPA) | Demand in Karnataka |
|---|---|---|---|---|---|
| QA / Test Engineer | Any degree + demonstrated rigour | Test planning, reproduction, automation | 3 – 6 | 12 – 22 | High — the most common first door |
| Gameplay Programmer | CS degree or strong portfolio | C#/C++, engine, maths, debugging | 6 – 12 | 25 – 50 | High |
| Engine / Systems Programmer | CS degree + low-level depth | C++, memory, rendering, profiling | 8 – 16 | 35 – 70+ | Moderate volume, very scarce supply |
| Network Programmer | CS + distributed systems interest | Replication, latency compensation, netcode | 8 – 15 | 30 – 60 | Scarce — high leverage for esports titles |
| Technical Artist | Art background + scripting | Shaders, Python, DCC tooling, optimisation | 6 – 12 | 25 – 48 | Very high — persistent shortage |
| 3D Artist / Animator | Portfolio-first; degree optional | Blender/Maya, PBR, rigging, animation | 4 – 9 | 18 – 35 | High — Karnataka's deepest existing pool |
| Game Designer | Shipped projects + written design | Systems, economy maths, documentation | 5 – 10 | 22 – 45 | Moderate — hardest to break into cold |
| Level Designer | Map portfolio, often from modding | Spatial design, engine tools, pacing | 5 – 10 | 20 – 40 | Moderate |
| UI / UX Designer | Design background + game literacy | Figma, engine UI, readability, accessibility | 5 – 10 | 20 – 40 | Moderate to high |
| Audio Designer | Audio production + implementation | Wwise/FMOD, sound design, mixing | 4 – 9 | 18 – 35 | Low volume, low competition |
| Build / DevOps Engineer | Software engineering background | CI/CD, Perforce, platform SDKs | 6 – 12 | 25 – 50 | Moderate, growing |
| Data / LiveOps Analyst | Stats, economics or CS | SQL, Python, cohort analysis, A/B testing | 6 – 12 | 25 – 50 | High — every live title needs this |
| Producer | Internal promotion or PM background | Scheduling, scope, risk, communication | 6 – 12 | 28 – 60 | Moderate |
| Esports Operations | Community, tournament or ops experience | Format design, logistics, refereeing, broadcast | 3 – 7 | 14 – 30 | Growing fast from a small base |
| Broadcast / Observer | Production or high-level play | OBS/vMix, observing, directing, casting | 3 – 7 | 12 – 28 | Growing — acute Kannada-language gap |
In almost every discipline here, a hiring manager will look at what you shipped before they look at where you studied. Three small finished things outperform one ambitious unfinished thing, every time.
QA, community management, localisation, esports operations and tools programming are all legitimate entries into studios — and internal movement between disciplines is far more common in games than in most industries.
Know what you are signing: IP assignment, non-competes, crunch expectations, revenue share definitions. A plain-language guide to first contracts is one of the first things the academy intends to publish, because it costs nothing to produce and saves people real money.
A whole software category exists around competition itself, and it is dramatically under-built in India. Much of it is buildable by a competent student.
Registration, seeding, bracket generation, check-in, dispute logging, results propagation. Any state-wide circuit needs this at scale, and there is no strong Indian-language option available today.
Ingest match data, compute performance metrics, generate coachable insight. Python, a data feed and good judgement about what actually predicts winning.
Overlays, live stat tickers, automated replay, multi-language audio routing, clip pipelines. Directly enables district events to look like real broadcasts.
Device fingerprinting, anomaly detection, invigilated remote play. High trust requirements, high impact, and an area where reputations are made.
One athlete, one verified identity across 31 districts, with guardian consent flows for minors and data-protection compliance built in from the start. This is the first thing the federation itself has to build.
Making competition work on 4G, on entry-level devices, in districts with intermittent connectivity. The hardest and most valuable problem on this list for Karnataka.
Karnataka's policy environment is genuinely supportive. Knowing the sequence saves years.
A finished 20-minute game with 500 real players teaches you more, and proves more to any funder, than a two-year unfinished epic. Every subsequent stage assumes you have done this.
Private limited or LLP, clean cap table, written founder agreement, IP assigned to the company from day one. Fixing this later is expensive; getting it right costs a few weeks.
Karnataka's AVGC-XR Policy 3.0 (2024–2029) provides incentives targeted at startups and MSMEs in animation, VFX, gaming, comics and XR. Registering early makes you eligible; registering late does not backdate.
State incubation infrastructure, the Gaming Centre of Excellence, and KITVEN Fund-4 (AVGC) — a dedicated venture fund for the sector — sit alongside the usual angel and accelerator routes.
Decide deliberately. A publisher buys reach and funding at the cost of control and revenue share. Self-publishing keeps both and demands that you build marketing capability yourself.
If the game works, the company becomes a LiveOps company. Plan for that org from the start: telemetry, community, moderation, seasonal content and a patch cadence you can actually sustain.
Everything below is free and runs on a modest machine. There is no financial barrier to starting — only a time one.
Pick one engine. Complete one official tutorial end to end without substituting your own ideas. Resist the urge to improve it.
Rebuild Flappy Bird or Breakout from scratch. You are learning the engine's shape, not designing.
Take your clone and alter a single mechanic. Observe how much else breaks. This is your first real design lesson.
Upload to itch.io with a title screen and credits. Get five strangers to play. Watch silently. Write down every hesitation.
Do that once and you are ahead of most people who say they want to make games, because you have finished something. Do it four times and you have a portfolio.
For engine, systems and network programming, a computer science background is a genuine advantage and often a filter. For art, animation, audio, design, QA, production, esports operations and broadcast, a portfolio of finished work matters far more than a degree. The most common route into an Indian studio without a games-specific degree is QA, followed by internal movement.
Partially. You can learn design theory, write and critique design documents, do pixel art on a phone, and study games analytically — all genuinely valuable. But engine work realistically needs a computer. Shared lab access is exactly the gap the planned academy is meant to close, which is why the Foundations curriculum is designed to run on 8 GB machines with integrated graphics. Until that exists, a college computer lab or a local gaming cafe is the practical route.
The honest answer: it is already changing which tasks are valuable, and the tasks it accelerates most are disproportionately entry-level ones — which is a real problem for people trying to get in. What has not been automated is judgement: knowing what to build, what to cut, when a mechanic is fun, and how to lead a team to finish. Our recommendation is to learn the tools seriously rather than avoid them, and to build a portfolio that demonstrates decisions, not just output.
For most students in Karnataka, Unity — because it dominates the domestic mobile studio market, which is where most local jobs are, and because C# is a gentler first language than C++. Learn Unreal if you specifically want high-fidelity 3D, shooters or virtual production, or if you are coming from the VFX and animation side where Unreal overlaps with existing skills. Godot is the best pure learning engine, and everything you learn there transfers.
That is the commitment: free tracks, free campus chapter charters, no affiliation fee — permanently. It is affordable specifically because the curriculum is designed around free software and existing college hardware rather than around new purchases. What cannot be provided free is a personal laptop, which is why shared lab access is central to the model. None of it is running yet, so nothing is being charged for either.
That is the design intent — evenings and weekends on a twelve-week cycle with a shipped artefact at the end, with academic-integrity guardrails on scheduling. We would rather someone take a track twice at a sustainable pace than burn out finishing it once.
Write to academy@kef.org.in. No chapter exists in any district yet — that is the current state everywhere, not a gap specific to you. The intended requirements are a faculty sponsor, institutional consent and fifteen student members, with no fee and no minimum infrastructure. Registering now puts your institution in the first group when it opens.
The academy isn't running yet — but everything in the starter kit above is free and available today. Finish one small thing. Then tell us you want a chapter in your college, and you'll be in the first group when it opens.