Varghese Chacko

Architected and launched the Docketry Suite — a multi-agent AI platform spanning ExtractIQ, NeuroDesk, and AI Flow — expanding client adoption across BFSI and enterprise domains. Built on DAG-based pipelines, node-level execution conditions, run tracking, and token/cost accounting: Django REST Framework and Next.js on Azure Kubernetes Service, with Celery, Redis, and PostgreSQL underneath. Diagnosed and fixed rolling-deployment failures, frontend asset drift across pods, Celery result-backend growth, and large-file streaming from Blob Storage. Cleared a 6-month document backlog and cut turnaround time from 3 months to hours, delivering $10.3M in customer savings; a newer release now processes 10M files a month at 2.5 minutes per file, down from 1 hour, with projected additional annual savings of roughly $15M. Projected to grow revenue from its first enterprise customer from $200K to $700K a month, with two additional prospective customers in the pipeline at roughly $250K a month each.

• Django REST Framework
• Next.js
• Celery
• PostgreSQL
• Redis
• AKS

Designed n8n-orchestrated workflow pipelines integrated with retrieval-augmented generation — connecting LLMs to internal knowledge bases and vector stores for automated document search, structured extraction, and task automation across engineering and support workflows.

• n8n
• RAG
• Vector Search
• LLM Orchestration

Bridged engineering, infrastructure, and production support for regulated financial-services systems — payment and recurring-card processing, funding bank files, Positive Pay, and overnight batch cycles. Automated BFSI reconciliation processes, cutting execution time from hours to minutes. Built and led L1/L2 support coverage across multiple time zones, from incident triage through API contract changes and release coordination.

• API Integrations
• Batch Processing
• Incident Response
• Release Governance

The document-intelligence engine behind the Docketry Suite — pipelines combining Amazon Textract, Azure Document Intelligence, and PaddleOCR with LLM-based structured extraction, including validation and retry handling for malformed model output, and large-file streaming to replace base64 payloads.

• Amazon Textract
• Azure Document Intelligence
• PaddleOCR
• LLM Extraction

Co-founded in 2016; served as Head of Technology & Principal Consultant through 2022. Scaled the engineering team from 8 to 25, led an enterprise cloud migration that cut infrastructure costs 25% YoY, compressed release cycles from 4 weeks to 1 week, and expanded a platform to 100K+ daily active users at 99.9% uptime — cutting downtime 35% with predictive AI monitoring. Took products from concept to deployment on limited resources — including Room Ring (later acquired by Roomi), Elity.co, and Eazy Gym, an IoT-enabled gym platform — handling architecture, engineering, and client delivery end to end.

• Product Architecture
• IoT
• Client Delivery