Aniruddha Chakravarty

About

Built and productionized a predictive maintenance system (predictEngineLife) estimating Remaining Useful Life for PW4000 turbofan engines from large-scale, noisy telemetry—emphasizing modular pipeline design, deterministic preprocessing, and strong observability/guardrails. Also has hands-on experience diagnosing multi-agent LLM customer-support workflows (schema/state issues, fallback paths, regression tests) and has led developer workshops (GDG Pune) while partnering with sales teams on technical discovery and POCs.

Experience

Education

Key Strengths

• Productionizing ML systems with scalable, loosely coupled pipeline stages
• Strong data quality engineering for noisy real-world telemetry (normalization, outliers, missing data, sensor drift)
• Designing extensible systems via configurable parameters and modular interfaces
• Building guardrails for predictable inference (sanity checks, bounded outputs, fallback behavior)
• Observability-driven operations (logging preprocessing stats, latency, predictions to detect regressions)
• Systematic real-time debugging of LLM/agent workflows by isolating failures across stages
• Fixing agent orchestration issues via stricter schemas, state validation, and regression tests from production traces
• Effective technical demos/workshops tailored to developer needs, including failure modes and operational concerns
• Partnering with sales/customer teams to de-risk evaluations with targeted demos/POCs and proactive handling of stability/security/ops concerns