sarthak biswas

ai engineer

(open for work)

i build what comes after pretraining — fine-tuning pipelines, rl environments, reward systems, and production ml that ships.

writes

experience

ai engineer

nov 2025 - present(full time, remote)

built the ai video generation pipeline from scratch — from llm orchestration and model finetuning to production apis serving thousands of users.

video generation pipeline

built an end-to-end pipeline that converts text/image queries into animated educational videos with multilingual narration in ~30 seconds. powered 280k+ videos across 14+ languages using manim for animation rendering and elevenlabs for text-to-speech.

llm orchestration

designed a multi-stage llm pipeline with model routing, tool calling, and async task execution. serves 45k+ active users with reliable throughput.

dataset engineering & finetuning

built sft/dpo/repair training data pipelines using unsloth ai for domain-specific model finetuning across multiple subjects — covering supervised fine-tuning, preference optimization, and error recovery.

rag & search

built a rag pipeline that indexes educational content into vector embeddings, retrieves relevant context at query time, and feeds it into the llm for grounded, accurate responses — reducing hallucinations and enabling subject-aware answers.

recommendation system

built a personalized video recommendation engine with multi-source candidate generation (topic-based, collaborative filtering, trending, exploration), embedding similarity search using sentence-transformers and pgvector, and a weighted ranking engine scoring on topic match, freshness, and popularity. handles cold-start users with trending/diverse fallback, and adapts feed composition as users warm up. deployed on kubernetes with 6 background workers for async feed generation.

fastapipostgresqlpgvectordockers3hugging faceunsloth aisentence-transformerskubernetes

projects

stock trader rl environmentapril, 2026
nyc eta enginemay, 2026
autonomous trader agentmarch, 2026

skills

/ AI / MACHINE LEARNING
PyTorchHugging FaceScikit-learnUnsloth AILLMsNLPReinforcement Learning
/ TRAINING / TECHNIQUES
SFTDPORAGPrompt EngineeringLLM Fine-tuningData Pipelines
/ DATA / LIBRARIES
PandasNumPyOpenCVNLTKMatplotlib
/ BACKEND / DATABASES
FastAPICeleryRedisPostgreSQLVector Databases
/ CLOUD / DEVOPS / MLOPS
AWS/GCPDigitalOceanDockervLLMCI/CDGitHub Actions
/ LANGUAGES & TOOLS
PythonTypeScriptGitLinuxVim