Sai Nikhil Kunapareddy

A summer at an early stage biotech startup building internal systems from scratch, which took as much systems thinking as coding and a lot of talking to the people who would actually use them. I built the pipelines that validate and process the datasets from their wet lab, an in-house end-to-end RAG system, and AI agents that let scientists analyse private datasets offline.

• Python
• FastAPI
• RAG
• Pinecone
• AI Agents
• Ollama
• Kafka
• Docker
• Spark
• GCP

I had wanted to do research since before the master's began, so I joined the Goodwill Computing Lab in my first year. I started out helping PhD students with automations, web scraping and building clean, reliable datasets, then ran a project of my own evaluating, benchmarking and fine-tuning LLMs as schedulers for high performance computing.

• Python
• Web Scraping
• LangGraph
• LangChain
• QLoRA
• Llama 4
• LLM-as-judge

The data science initiatives at ZF convinced me this was what I wanted to do long term, so I moved to the US to study it properly. Two years of coursework across the full machine learning lifecycle, from data management through to deployment.

• Algorithms
• Data Management
• Data Processing
• Supervised ML
• Unsupervised ML
• Deep Learning
• LLMs
• MLOps

After the promotion I joined ZF's data science initiatives, generating simulation data to train surrogate models and building pilot models from scratch for the internal pain points where machine learning looked like the answer. I also engineered the ML inference services, handling 1K+ concurrent requests at sub-second latency.

• Python
• Scikit-learn
• XGBoost
• SHAP
• SageMaker
• CloudWatch
• TypeScript
• Spring Boot
• AWS