Janakiraman Giridhar Sivasailam

AI/ML engineering leader with over 10 years of experience in developing and deploying production-grade ML platforms across fintech, healthcare, and retail sectors. Achieved 33% automation and 92% model accuracy by delivering scalable AI solutions that process millions of records. Collaborated with cross-functional teams to define requirements and measure business impact, resulting in significant MTTR reduction and cost savings. Holds an executive PG diploma in management and AI from IIM Indore, with a focus on product strategy and agile methodologies, transitioning to AI product management to drive vision and roadmaps for ML platforms.

• Digital Employee Sentiment Intelligence Platform (Ivanti, 2024-Present): Built full-stack GenAI solution for summarizing unstructured ITSM/EXM data → p95 latency ~1.5s → cost $0.005/request → eval precision/recall 0.85 (offline benchmarks/synthetic tests) → GPT-4o + RAG + LangChain + PineconeDB + Power BI → Ops: Azure/Docker, CI/CD (MLflow), drift monitoring/alerts → Reduced manual triage 28% via caching/chunking optimizations; postmortem: Addressed hallucinations by grounding in enterprise docs, improving confidence 30%.
• ITSM Ticket Auto-Triage & Resolution (Ivanti, 2021-2023): Developed multilingual classification system for automating triage/routing/resolution → p95 inference
• Clinical Trial Phase Transition Prediction (Nous, 2019-2020): Created predictive models for go/no-go R&D decisions → eval AUC 0.92 → cost $0.01/query → precision/recall 0.88 (cross-validation) → Spark-MLlib + Python + Databricks → Ops: REST APIs, monitoring notebooks → Improved decision speed 3x; postmortem: Handled imbalanced data with oversampling, raising minority class recall 15%.
• Payment Fraud Detection PoC (Personal/Ivanti-Inspired, 2023): Built real-time anomaly detection pipeline → p95 ~150ms → cost $0.004/transaction → eval F1-score 0.90 → XGBoost + Flask + Redis + Kafka → Ops: Docker/CI, alerts for degradation → Reduced false positives 47%; postmortem: Optimized I/O bottlenecks with stream processing, cutting latency 25%.
• Smart Vision-Based Vending Machine PoC (Personal, 2022): Designed touchless purchase system → p95 checkout 1.7s → cost $0.003/transaction → detection accuracy 96.4% → YOLO + Raspberry Pi/Nvidia Jetson + MongoDB + Flask APIs → Ops: Edge deployment, live pilots → Increased user satisfaction 78%; postmortem: Mitigated low-light errors with data augmentation, improving robustness 20%.
• ReDi DW Migration (Nous, 2017-2021): Migrated legacy to scalable DW → processing time reduced 30% → query performance +40% → Hadoop + Spark + Hive + Kafka + Snowflake + NiFi → Ops: Lambda arch, CI/CD, data governance checks → Enabled real-time reporting; postmortem: Resolved integration delays via impediment removal, ensuring on-time delivery.