Hi, I'm Nishchay.
Data Scientist & ML Engineer.
Co-founder at ContextJet.ai.
M.S. Data Science, UC San Diego ('25–'27)Applied Gen AI & LLM SystemsCo-founder, ContextJet.aiEx-C5i.ai, Senior AnalystBased in CA
What I build
AI/LLM Agents
Task agents for ops/support. Prompt + tool + eval.
Retrieval / GraphRAG
Knowledge-aware answers over internal data.
Automations for analytics teams
Pipelines, guardrails, logging.
Selected builds
Generative AI frameworks for tech & CPG
Delivered multiple GenAI frameworks that helped drive $500K+ in revenue; focused on prompt/weight optimization for reliability.
Role: design, orchestration, deployment
LLM prompt/weight optimization system
Designed scalable optimization framework that improved performance by ~60% and made rollouts reproducible.
Role: experimentation, MLOps
Data ingestion & hybrid modeling pipeline
Automated ingestion + QA using Bayesia, improving processing efficiency by ~40%.
Role: pipeline build
In-house LLM for rapid POCs
Enabled 5+ POCs across domains, cutting model delivery time by 35%.
Role: model setup, productization
Writing & thinking
Articles
Notes
Agent UX is mostly constraints.
Eval before you scale.
Ship small, wire data after.
RAG should survive noisy queries.
Current
- —Building client-facing agent workflows at ContextJet.
- —Experimenting with retrieval evals and guardrails.
- —Documenting GenAI survey work for follow-up post.
- —Balancing UCSD coursework with studio builds.
Signals
- •Star Innovator & Ideathon/Buildathon wins at C5i.ai (2023, 2024)
- •Rookie of the Year, 2023
- •Speaker at MLDS 2024 (Lattice)
- •Runner-up, Techniche Hackathon, IIT Guwahati