Venkateshh Sugandham

ML/GenAI Engineer with hands-on experience building production time-series anomaly-detection and root-cause-analysis systems, plus GenAI applications including RAG and multi-agent systems.

Skilled across the full stack: data engineering (streaming pipelines, time-series databases), deep learning (autoencoders, forecasting), and LLM tooling. Proficient in Python, SQL, LangChain, and vector/graph databases.

Intern — ICT Services, Wind Predictive Maintenance & Anomaly Detection - Greenko Group - Hyderabad

(2026-04)

• Built a per-turbine MLP autoencoder Normal Behaviour Model for early-warning anomaly detection, achieving 67.4% episode detection and ~36-hour median lead time across 48 turbines in the Axis plant pilot.
• Benchmarked the production model against PCA, GRU, and zero-shot foundation-model baselines under a standardized held-out evaluation protocol spanning 1,156 alarm episodes.
• Designed a fleet-scale, multi-OEM architecture combining TimescaleDB (SCADA telemetry), Neo4j (knowledge graph), and an on-prem LLM for governed root-cause diagnosis narration.
• Built the production ingestion pipeline (MQTT → Telegraf → TimescaleDB) with ReplacingMergeTree-style deduplication, partitioning, and tiered TTL/cold-storage strategy.
• Contributed to migrating a legacy Java/Struts/JSP ITAM module to a Node.js/Express/React stack, preserving full feature parity against the existing MySQL database.

Master of Science - ML & AI - Liverpool John Moores University, UK (Online, upGrad) (2025)

B.Tech - Electrical Engineering - Indian Institute of Technology Jammu (2017 - 2021)