Senior AI Systems Architect with 18+ years spanning three career phases: enterprise backend engineering at scale, lead development across concurrent fintech and logistics engagements, and production AI/LLM systems architecture. Built document intelligence pipelines achieving 97% extraction accuracy and reducing KYC review by 65% in regulated production environments.
Specialist in Kafka-native real-time AI inference, multi-stage LLM orchestration, and agentic workflow design — with full observability from day one. NVIDIA Agentic AI certified. Open to sponsorship and relocation globally.
MeasureOne Inc.
97% data extraction accuracy on heterogeneous financial documents — multi-stage pipeline: deterministic parsers + LLM fallback inference, automated RAGAS-style eval coverage across 10+ document categories. 65% reduction in manual KYC document review — production document intelligence pipeline (Python, Node. js, AWS Textract, Kubernetes) processing compliance documents at fintech scale.
Reduced data latency from daily batches to minutes — near-real-time CDC pipelines with full Grafana + Datadog observability across extraction runs.
SJULTRA Inc.
Remote, US
30% throughput improvement — event-driven microservices with Kafka, Mongo DB, and LLM inference layers; owned schema → API → deployment → metrics end-to-end. 60% reduction in compliance review time — LLM-based regulatory document automation with audit trails integrated into reporting systems.
Independent Consultant
Predictive financial data pipelines for document intelligence clients; LLM observability dashboards tracking pipeline accuracy, model drift, and anomalies.
SS Shipping Pvt Ltd
60% reduction in customs processing time — automated end-to-end documentation and shipment workflows; real-time fleet tracking with predictive maintenance alerts reducing breakdowns by 20%. Mongo DB read-optimised data layer with denormalized/materialized views for low-latency logistics APIs handling high-throughput fleet operations.
Real-time Kafka data pipelines for logistics event streaming — event-driven architecture supporting fleet, shipment, and compliance data flows.
Ekshot Advertcom Pvt Ltd
40% campaign efficiency improvement — multi-tenant marketing platform with automated campaign analytics and real-time Kafka pipelines. 25% user acquisition increase — growth automation integrations (Mailchimp, Stripe, Zapier); 50% backend performance uplift via legacy module refactoring.
TEKsystems / MasterCard
Pune, India
Backend modules for card upgrade and migration systems (Java, Apache Struts) at large-scale card lifecycle; secure payment APIs reducing code duplication by 30%; IBM Web Sphere cluster deployments; batch-driven enterprise billing reporting.
IBM
Enterprise backend engineering, ETL pipelines, compliance systems, and distributed data processing across financial services, media, and logistics domains.
UTV
Enterprise backend engineering, ETL pipelines, compliance systems, and distributed data processing across financial services, media, and logistics domains.
LRN
Enterprise backend engineering, ETL pipelines, compliance systems, and distributed data processing across financial services, media, and logistics domains.
Global Eagle
Enterprise backend engineering, ETL pipelines, compliance systems, and distributed data processing across financial services, media, and logistics domains.
JNEC
Enterprise backend engineering, ETL pipelines, compliance systems, and distributed data processing across financial services, media, and logistics domains.
B.E. Computer Science
First Class with Distinction
2023 – Present
Engagements: Measure One Inc. (Lead AI Engineer, 2024–2025), SJULTRA Inc. (Lead Engineer, Remote US, 2023–2024), Independent Consultant (2025–Present)
2017 – 2023
Engagements: SS Shipping Pvt Ltd (Lead Developer, 2017–2022), Ekshot Advertcom Pvt Ltd (Lead Developer, 2021–2023) [concurrent]
2007 – 2016
Engagements: TEKsystems / Master Card (Senior Developer, Pune, 2015–2016), IBM, UTV, LRN, Global Eagle, JNEC (2007–2015)
Stack: Python, Lang Graph, FastAPI, pgvector, Lang Smith, AWS. Lang Graph multi-agent pipeline: PDF ingestion → LLM extraction → structured output → Postgre SQL. RAGAS evaluation harness + Lang Smith tracing. Deployed AWS ECS.
Stack: Python, Kafka, Lang Chain, Grafana, Docker. Kafka consumer → LLM classification/summarisation → Kafka producer. Sub-second latency. Grafana dashboard: throughput, latency, model drift.
Training
Digital Natives
In progress
Targeting Q3 2026
18+ years of production engineering discipline across concurrent senior engagements — a career pattern common at architect and staff level globally. Every AI system I build is observable, monitored, and fault-tolerant from day one. My Kafka background enables real-time LLM inference pipeline design that most AI engineers cannot replicate.
Consistent delivery in regulated, high-stakes environments: fintech, KYC, compliance, logistics. Available for global relocation or full-time remote within 2 weeks.