"Every great engineer thinks 10 moves ahead."
In chess, you develop your pieces before you attack. In engineering, you develop yourself before you build.
I'm Adarsh Kumar Chaurasiya — SDE-1 at Rupeek Fintech, Bengaluru. I treat every system like a chess position: study the board, calculate the lines, and only then make the move.
I play on two boards at once. On one, I build Conversational AI — real-time voice agents with Speech-to-Text, LLM reasoning, and Text-to-Speech over Exotel, orchestrated with LangGraph. On the other, backend engineering in Node.js and Python — Zeebe/Camunda workflow workers, cron-based automation, and a high-availability eSign platform with circuit-breaker failover that I architected from scratch.
Before this: B.Tech in IT from G.L. Bajaj (CGPA 8.6, Dept. Rank 3), 1000+ LeetCode problems, and a stint mentoring 600+ students in DSA & ML at Coding Ninjas. Engineering, like chess, isn't about memorizing moves — it's about understanding positions.
You castle to secure your king before the battle begins. These are the skills my career is castled behind.
Backend services, REST APIs, workflow workers & automation — my main battery.
FastAPI, AI agents, data pipelines — moves in every direction, like a queen.
1000+ problems of pure calculation training. Tactics win games.
LangGraph agents, LLMs, prompt engineering, STT/TTS voice bots, LiveKit.
MongoDB, MySQL, SQLite — every position stored, nothing forgotten.
Circuit breakers, failover strategies, HLD/LLD — king safety for production.
The middlegame is where plans become attacks. Every project below is a piece I've played — each with its own role on the board.
The flagship — a multi-lender, multi-provider document-signing platform with an Opossum circuit-breaker failover system I architected from scratch. It monitors provider health, auto-switches traffic when failures spike (with Redis-backed state and Flagsmith feature flags), and recovers when the primary stabilizes. Eliminated a major failure point in the signing journey. Lose this one, lose the game.
A real-time voice agent for Release-out-Bond calls — Speech-to-Text, LLM reasoning, and Text-to-Speech over Exotel for human-like conversations, plus intelligent slot-booking automation driven by customer intent and consent.
IG & Consumption Loan integrations with scheme-based credit assessment for personalized recommendations — plus Zeebe workflow workers and cron jobs that generate and ship renewal files to FED, reliably, on schedule. Straight-line power.
"What movie should I watch?" — answered. PickAMovieForMe ends the endless Netflix scrolling: NLP CountVectorizer + cosine similarity over Bollywood & Hollywood titles picks your next film in seconds. Try it live ↗ · Read the blog ↗
Sees the pattern across the whole board — OTP verification fused with geolocation tracking; transactions only succeed if the initiator is within 200 m of the cardholder. Fraud spotted on the long diagonal.
Regression and NLP models, Tableau dashboards, web scrapers with BeautifulSoup & Selenium, Streamlit apps. Every pawn dreams of promotion — some of these already have.
A grandmaster's strength shows in the game log. Here's mine — every move annotated.
Agentic AI voicebot for Release-out-Bond calls (STT + LLM + TTS over Exotel), IG & Consumption Loan integrations with scheme-based credit assessment, Zeebe workflow workers, cron-driven FED renewal automation, and the circuit-breaker eSign failover system. Strong central control of the board.
Built an agentic chatbot from scratch with LangGraph — dynamic tool selection, custom API wrappers — and the document eSign orchestration service that internal teams call to trigger signing, track status, and fetch secure signing URLs.
Resolved ~600 doubts at a 4.73/5 rating across Data Structures, Algorithms and ML, and evaluated dozens of machine-learning projects. Teaching a line is the fastest way to truly understand it.
Built and shipped models on Linear Regression and worked on Natural Language Processing — the first practical conversions of theory into position.
The opening move. CGPA 8.6, Department Rank 3 — data structures, algorithms, DBMS, and the first time code felt like calculating a combination.
In the endgame, only what you've truly earned remains on the board. These are mine.
Winner — GIH Hackathon · Runner-up — Nest Hackathon (Regional).
Dept. Rank 3 · CGPA 8.6 — B.Tech IT, GLBITM '25.
TCS CodeVita S11 — Global Rank 3520 · Intercollege Coding Competition — Rank 4.
600+ doubts resolved at Coding Ninjas · 4.73/5 rating in DSA & ML.
Yes — I actually play. Challenge me to a game along with the interview.
The game is won — let's set up the next board together.