Headline

GenAI & AI-Platform Leader | Agentic Workflows, RAG, LLM Evaluation & MLOps at Scale | Technical Program Manager @ Reliance Intelligence | Ex-EXL (Media Vertical), Optum | Databricks Certified | IIT-KGP · IIM-L

About

I build production GenAI systems — and the platforms and teams that keep them running. Over 9+ years across media, telecom, and healthcare, I've worked at the intersection of applied AI, data-platform engineering, and program leadership: taking AI from slide-deck pilots to monitored, production-grade systems with measurable business outcomes. Today, as a Technical Program Manager at Reliance Intelligence (Consumer Intelligence), I'm helping build an AI-powered consumer product — agentic orchestration, memory & personalization, and voice/multimodal experiences — coordinating data, personalization, and product-metrics workstreams across a dozen-plus delivery pods. Before that, at EXL, I led the AI engagement for a major US OTT streaming account, where I designed and led an enterprise Agentic AI Data Quality Platform — multi-agent workflows and RAG pipelines (BigQuery VectorStore + Vertex AI embeddings, semantic retrieval, RAGAS evaluation) that autonomously detect anomalies and cut ticket-resolution time by ~50%. I also ran rigorous LLM evaluations (text-to-SQL benchmarking across GPT-4, Gemini, Llama 3 and specialized SQL models) and scaled the MLOps backbone — 300+ Airflow DAGs on GKE moving 1TB+/day, 15+ production ML models on Vertex AI. What I focus on: • GenAI agents, LLM orchestration, RAG & retrieval-based QA, agentic workflows • LLM evaluation & LLMOps (benchmarking, RAGAS, prompt versioning, multi-model routing) • Scalable ML & data platforms (Vertex AI, Airflow, GKE, Databricks, Spark, BigQuery, Snowflake) • "Pilot → production" AI transformation and MLOps at scale • Technical program leadership, stakeholder governance, and team building I bring an engineer's depth (IIT Kharagpur) with a strategist's lens (IIM Lucknow) — Databricks Certified ML Associate — to turn complex AI ambitions into durable, working systems.

Location

Greater Bengaluru Area

AI stack

aistudio

anthropic apis

claude code

claude

codex

deepseek

Work experience

Technical Program Manager at Reliance Intelligence, Mar 2026 - Aug 2026 · 5 mo, Bengaluru, Technical Program Manager in the Consumer Intelligence org, helping build an AI-powered consumer application spanning agentic orchestration, memory & personalization, and voice/multimodal experiences. Own data, personalization, and product-metrics workstreams across a dozen-plus delivery pods. • Co-designed a Personalization Engine on an enterprise customer-profile platform — a layered architecture (signal ingestion → profile enrichment → candidate generation → low-latency serving → feedback loop) built for sub-200ms serving. • Stood up the program's metrics & instrumentation framework, with full lineage from each metric back to its source event — making product dashboards trustworthy and traceable. • Designing an Agentic SDLC — an agent-driven program-management layer (bug-tracing, spec-compliance, release-readiness, and status-rollup agents) to automate routine engineering workflow. • Authored the cross-cloud integration architecture connecting the app to an enterprise customer-profile platform, defining data planes, latency SLOs, and the authentication stack. • Drive delivery governance, release management (dev→SIT→prod), and third-party AI/observability vendor evaluation across the program.

Engagement Manager at EXL, Feb 2024 - Jan 2026 · 2 yrs, Bengaluru, Engagement Manager leading GenAI innovation and delivery for a major US OTT streaming account. GenAI Architecture & Builder Leadership: Designed and led end-to-end development of an enterprise-grade Agentic AI Data Quality Platform — multi-agent workflows (QA, SQL, ticketing, incident, monitor) and RAG pipelines (BigQuery VectorStore + Vertex AI embeddings, semantic chunking, MMR & conversation-aware retrieval, RAGAS evaluation) — enabling autonomous anomaly detection and reducing ticket resolution time by 50%. LLM Evaluation & Conversational BI: Built a natural-language-to-SQL-to-visualization BI system and ran a rigorous text-to-SQL benchmark across 200 questions on a novel schema, evaluating GPT-4, Gemini, Llama 3, Mistral, and specialized SQL models to select the production model. Commercial Strategy & Account Growth: Owned business-case creation and ROI modeling to secure internal investment — pitching a $500K R&D initiative for GenAI adoption that drove vertical expansion and a 30% revenue uplift within 9 months; secured two new verticals contributing ~20% further uplift. Executive Stakeholder Management: Managed C-Suite and VP relationships via quarterly SteerCos and data-driven QBRs, translating technical constraints into value narratives, achieving 95%+ CSAT and securing renewals. Program Governance & Risk Management: Orchestrated 5+ concurrent global workstreams with RAID logs, quality gates, and release planning, sustaining consistent on-time delivery across distributed regions. Platform Engineering: Directed the Clean Room migration from BigQuery to Snowflake (Snowpipe-based ELT, privacy governance) for secure, compliant integration of 1st- and 3rd-party data.

Project Manager at EXL, Jan 2022 - Feb 2024 · 2 yrs 2 mos, Bangalore Urban, MLOps Strategy & Pilot-to-Production Scaling: Defined and enforced the ML SDLC and built a reusable pilot-to-production framework; steered the lifecycle of 15+ production models (CLV, Reactivation, Habituation, Churn) on Vertex AI and MLflow with champion-challenger retraining and data-drift monitoring (KS, PSI), improving campaign/engagement metrics by 20%. Data Platform Engineering: Architected multi-tenant Airflow on GKE across 300+ DAGs processing 1TB+/day (Kubernetes Executor, SSO, Helm, DAG-level permissions), improving SLA adherence to 90% and platform reliability to 95%+ while reducing operational overhead by 30%. ML Data Quality Framework: Built a cloud-agnostic framework across six data-quality dimensions, monitoring 200+ tables with 2,000+ daily automated checks and 100+ ML models (Isolation Forest + SHAP, Prophet, SynapseML). Org Building & People Leadership: Led a 14-member cross-functional team (Data Engineering, MLE, Full-Stack) with competency matrices, structured onboarding, and career coaching — achieving 95% retention and a 30% reduction in hiring cycle time. Security & Automation: Engineered a self-service IAM provisioning system (Python/FastAPI) with a LangChain natural-language-to-SQL layer, automating access workflows across GCP/AWS and reducing turnaround time by 40%.

Assistant Project Manager at EXL, Jun 2020 - Jan 2022 · 1 yr 8 mos, Bengaluru, Pipeline Modernization & Cloud Transformation: Orchestrated the migration of legacy on-premise ETL workflows to a cloud-native stack (GCP + Airflow on Kubernetes), optimizing DAG performance to reduce data latency by 25% with zero downtime. Streaming Data Engineering: Built streaming pipelines (Apache Beam/Dataflow, Pub/Sub, BigQuery, Bigtable) moving data from the source platform to a CDP and downstream activation systems. Operational Excellence: Developed Python-based monitoring frameworks providing real-time pipeline-health visibility, ensuring a 95%+ job success rate. Stakeholder Enablement: Translated complex data insights into actionable recommendations for senior leadership, contributing to a 10% improvement in productivity.

Program Manager Intern at Amazon, Apr 2019 - May 2019 · 2 mos, Hyderabad Area, India, Program Manager Intern in Operations / Transportation Analytics, optimizing packaging and unloading efficiency across Amazon India's transportation network. Network Analytics: Analyzed 37.1M+ packages to identify Bagging Potential Fluids using a 3-flag classification algorithm (volume, dimension, weight) combined with box-barcode logic. Impact Quantification: Established a strong positive correlation (R² = 0.85 across 500 delivery stations) between fluid ratio and average unload time, and derived per-package unload times (bag 0.91s vs fluid 4.07s) to quantify ~1,100 hours/week of unloading-time savings, plus a network-wide reduction in package damage. Data Quality: Extracted and cleaned data from the India transportation database, reducing a 3% data-inconsistency rate to under 1% by cross-referencing tables, validated at a 98% confidence interval. Dashboarding & Rollout: Built an automated daily dashboard (day-on-day fluid view, top-5 defaulting origin nodes) for pan-India Ops rollout, driving action items with Ops and ACES leadership — node audits, revised bagging guidelines, and associate training.

Software Engineer at UnitedHealth Group, Jul 2015 - May 2018 · 2 yrs 11 mos, Bengaluru Area, India, Big Data Engineering: Migrated legacy MapReduce systems to an optimized Spark/Scala stack, improving performance by 500%; designed Spark-Scala algorithms for unstructured healthcare data, enhancing usability by 35%. Data Platform & ETL: Engineered the data-acquisition (ETL) layer pulling pharmacy and claims data from 9 source systems into a Hadoop data platform; built the ingestion pipeline (inbound landing → workflow orchestration → raw zone → Avro conversion → curated store) and the Hive processing and publication jobs exposed via JDBC for downstream analytics. Process Transformation & Automation: Delivered real-time reporting applications, reducing manual effort by 60% and improving SLA compliance to 90%; analyzed 100GB+ weekly HIPAA/PHI-regulated healthcare datasets, improving on-time SLA adherence by 15% YoY. Storage Optimization: Developed compression and columnar storage solutions (Avro, ORC, Parquet), cutting storage costs by 20% and accelerating query performance. Stack: Hadoop, HDFS, Hive (HQL, partitioning, SerDes), MapReduce, Spark, Scala, YARN, Sqoop, Java, SQL, Bash. Fast-track promotion within 8 months.

Summer Intern at Bhabha Atomic Research Centre, May 2013 - Jul 2013 · 3 mos, Mumbai Area, India

Education

Master of Business Administration - MBA, Indian Institute of Management, Lucknow

Bachelor of Technology (BTech) + Master of Technology (MTech), Electrical, Electronics and Communications Engineering, Indian Institute of Technology, Kharagpur

Public links

github: https://github.com/nikiit007