Open to Opportunities

Subarna Saikia

I build backend systems that think.

GenAI · Distributed Systems · Developer Tooling

SCROLL
01. about

About Me

I wrote my first line of code at 17. I haven't stopped since.

It started with competitive programming — late nights on CodeChef and Codeforces, learning to think in algorithms before I even knew what a framework was. When I joined my B.Tech program in 2022, I didn't rush into web dev tutorials like most. Instead, I went lower. Much lower.

I built Space Invaders. Then Snake. Then a fully functional Chess Engine in C++ with bitboard representation and an AI bot — all from scratch. That phase taught me something most developers never learn: how machines actually think.

From there, the trajectory was natural — backend systems, distributed architecture, and eventually the intersection I'm most excited about: Generative AI meets production engineering. Today I architect pipelines where LLMs aren't just demos — they're load-bearing components of real systems.

I'm a systems thinker first, an engineer second, and a problem-solver always.

engineer.py
1subarna = Engineer(
2core = ["Backend", "GenAI", "Systems"],
3weapons = ["Python", "Java", "C++", "Go"],
4current = "Building AI-native pipelines @ Pocket FM",
5rating = {"CF": 1432, "CC": 1670},
6kaggle = "Top 2% Globally 🏆",
7status = open_to_work(),
8)
02. experience

Where I've Worked

P

Pocket FM

Software Engineer Intern

January 2026 – Present

Bengaluru, India

  • Architected an end-to-end automated data ingestion pipeline using WhisperX and Pyannote-Audio — generating speaker-diarized transcripts with word-level timestamp precision for large-scale audio content.
  • Built scalable web crawlers for automated multimedia data acquisition, eliminating ~60% of manual data curation effort and unlocking faster model training cycles.
PythonWhisperXPyannoteNLPGenAIData Engineering
A

Automatic Data Processing (ADP)

SDE Intern

July 2025 – September 2025

Hyderabad, India

  • Designed an event-driven consistency mechanism using Spring Boot + Apache Kafka to synchronize deletions across payroll worksheets and time-off records — zero data drift across services.
  • Built a high-throughput async consumer service for data purging that achieved 100% data integrity across distributed microservices.
  • Optimized multi-level SQL joins to detect and eliminate orphaned records, cutting query execution time by ~40%.
JavaSpring BootApache KafkaSQLMicroservicesDistributed Systems
L

The LNM Institute of Information Technology

Undergraduate Research Intern

June 2025 – July 2025

Remote

  • Conducted research on Protein Secondary Structure Prediction using deep learning — benchmarked BiLSTM and DeepCNF architectures against traditional ML baselines.
  • Achieved Q3 Accuracy of 91.55% through systematic hyperparameter tuning on complex biological sequence datasets.
PythonBiLSTMDeepCNFDeep LearningResearchBioinformatics
03. projects

Things I've Built

A selection of projects across backend systems, AI/ML, and developer tooling.

🧬
Full-Stack · NLP Platform

Genesis

A production-grade NLP annotation platform for coreference resolution — built as a modular monolith with enterprise-level architecture.

JavaSpring BootOAuth2JWTReactDockerMaven
📝
Full-Stack · AI-Powered

NoteX

An AI-native note-taking app that doesn't just store your notes — it understands them. Semantic search, auto-revision, and dynamic quizzes powered by LangChain + OpenAI.

Next.jsNode.jsTypeScriptMongoDBLangChainOpenAIRAG
☀️
AI/ML · Hackathon

Solar Efficiency Prediction

Ranked Top 25 globally (out of 1500+ teams) in the Zelestra × AWS ML Ascend Challenge. An ensemble ML pipeline for predicting solar panel efficiency from noisy sensor data.

PythonCatBoostEnsemble LearningMICEFeature Engineering
♟️
Systems · C++

Chess Engine

A fully functional chess engine built from the ground up in C++ — bitboard representation, SDL2 GUI, and an evolving AI bot. The project that proved I could build anything from scratch.

C++SDL2BitboardsAIGame DevOOP
06. contact

Get In Touch

Let's build something remarkable.

Whether you have a role, a project, or just want to talk systems and AI — my inbox is always open.

Typically reply within 24 hours.