Experience · Gurgaon, India · May—Jul 2026
Windflow.ai
AI Engineering Intern
- Shipped a production LLM voice interviewer on GCP Cloud Run probing agents on 4 diagnostic pillars.
- Unified 14 OCR engines and 9 LLM extractors into a single benchmark suite with 250+ automated tests.
- Evaluated a 20-model forecasting zoo on 12 public datasets; GBMs beat seasonal-naive by 9–18% on RMSSE.
- Led a pre-registered A/B study on 5 paired builds; measured 4–5% token savings, advised against adoption.
- Developed a human-in-the-loop golden pipeline: ROVER voting, Dawid–Skene EM, and LLM arbitration.
- Instrumented per-call LLM cost analytics (Langfuse, LiteLLM) on a PostgreSQL backend with abuse detection.