Member of Technical Staff, Applied AI
GC AI is building the legal AI platform that 1,800+ companies rely on. I have been building the same multi-model LLM orchestration, RAG pipelines, and eval-gated production AI for 7+ years. FacadeDriver routes 30+ foundation models. LLM-as-Judge gates block production. 1,690 ground-truth samples.
Relevant Work Experience Explore All ›
Open Source Explore All ›
I'm the engineer your ML platform team calls when the foundation model demo worked once and now production is on fire.
My day-to-day at Airbnb is the unglamorous half of AI: multi-model isolation, routing across 30+ LLMs, retry and fallback when a provider rate-limits you at 2 AM, PII detection that doesn't false-positive on the word "John," and evaluation frameworks that catch drift before users do.
I've spent 7+ years migrating between stacks - Java at Oracle, PySpark at Shell, Kafka at Southwest, Spring Boot & Kubernetes at Eli Lilly, and now Python / FastAPI / Airflow / AWS Bedrock at Airbnb. What persists is a taste for systems that fail gracefully, observe themselves, and don't wake humans at 3 AM.
GC AI is building the legal AI platform that 1,800+ companies rely on. They need multi-model LLM orchestration, legal-grade evals, RAG pipelines for complex document analysis, and rapid prototyping frameworks. That is exactly what I build at Airbnb today. I want to bring this to GC AI.
Multi-tenant ML platform
Built Airbnb's BPI Virtual Analyst - 4 partner teams, 128+ concurrent internal devs, 30+ LLMs behind one interface. Provider lock-in is a config flag.
16x scale on workflows
Refactored the analysis pipeline from 600 rows to 10,000 per run. Same hardware, batching & chunked uploads.
23-version eval harness
LLM-as-Judge eval gates blocking production when precision drops below 0.85. 1,690 versioned ground-truth samples.
System Design Architecture
Production architectures I built at Airbnb, mapped to GC AI's product surface
A tour through five production stacks.
Five companies, four industries, one common thread: production-grade systems that data scientists trust.
- 2024 - NowContractAirbnb - San Francisco
Senior Software Engineer - Gen AI Development Experience
GenAI tooling, Labelbox workflows, PII pipelines, Airflow. 30+ LLMs, 128+ concurrent internal devs, 10K rows/run.
- 20247 mosEli Lilly - Philadelphia
Senior Software Engineer - Dose Management System
Dose order management for radiopharmaceuticals on KUBED. Spring Boot, React, Temporal, Istio. 99.9% uptime.
- 2023 - 241 yr 1 moSouthwest Airlines - Dallas
Senior Software Engineer / Data Scientist
Kafka streams for real-time flight tracking. RabbitMQ microservices. 95% test coverage via Gradle / Java port.
- 2021 - 221 yr 11 mosShell PLC - Houston
Senior Software Engineer / Data Scientist
NLP text-classification for drilling-loss events. CNN autoencoders for acoustic impedance. Responsible-AI POC with LIME / SHAP.
- 2017 - 192 yrsOracle India - Bangalore
Software Engineer
Cloud HCM integrations, 50K+ LOC. ERP analytics for 13 business units. ELK + Kinesis fraud-detection POCs.
The stackI've shipped with.
Tools I'd reach for tomorrow, ordered by how often they ship in production at my last 5 jobs.
What colleagues say
He doesn't just deliver - he continuously looks for ways to make things better.
Sai has been an outstanding partner in the deployment of the BPIVA tool, and I want to take a moment to recognize his incredible contributions. Thanks to Sai's efforts, the BPIVA tool has had a significant impact on reducing non-value-added work, enabling the BPI team to shift their focus to high-impact, actionable tasks exactly where their energy should be. What truly sets Sai apart is his deep understanding of technology combined with his ability to quickly grasp tool requirements and translate them into real solutions. He doesn't just deliver, he continuously looks for ways to enhance and upgrade the tool's capabilities, ensuring it evolves alongside our team's needs. None of this would have been possible without Sai's dedication and expertise. He is a truly great partner who consistently goes above and beyond to deliver excellence. Thank you, Sai, for everything you do! 🙌
Ameet Shinde
Senior Manager, BPI · Airbnb
GC AI is building
what I've already built.
Your Applied AI team is building multi-model LLM orchestration (5 models tuned per task), legal-grade evals that harden prompts and retrieval, RAG pipelines for complex document analysis, and rapid prototyping frameworks for weekly model releases. That is exactly what I do at Airbnb today. I own the BPI Virtual Analyst - a GenAI platform abstracting 30+ foundation models behind FacadeDriver with routing, retry, fallback, and graceful degradation, serving 128+ users across 4 partner teams. I built the 23-version LLM evaluation harness with 1,690 ground-truth samples and LLM-as-Judge gates that block production when precision drops below 0.85. I built end-to-end RAG pipelines with reranking and contextual compression, multi-agent systems with tool calling and agentic memory, and rapid prototyping frameworks consumed by other teams. Your Exact Quote system demands verbatim accuracy on every citation. My eval gates demand the same. I want to bring this to GC AI.
Where I learned to do it.
Let's talk
GC AI.
Open to Applied AI roles at GC AI
SE 4/5 - Remote (Houston, TX) - H-1B transfer (I-140 approved, AC21 portable). Available immediately.
Start a conversation