Gaurav Dombale

Software Engineer with over 2+ years of experience specializing in Full Stack application's and AI engineering. Shipped 5+ production applications including multi-agent LLM systems and enterprise SaaS platforms.

AI Engineer at SoluLab (2026-06 – Present)

ConX AI (NestJS, Node.js, Python, FastAPI, PostgreSQL, pgvector, Redis, OpenAI, WebRTC, Twilio, React, Docker, AWS)

• Architected a platform-wide fraud detection system across 5 microservices, combining regex, pgvector similarity, and LLM-based detection to identify prompt injection and jailbreak attacks while reducing LLM usage by ~95%.
• Built a 4-layer behavioral fraud engine with Redis, HyperLogLog, EWMA/z-score anomaly detection, analyst feedback, and automated model retraining.
• Developed a multi-tenant AI usage and cost tracking system for LLM, STT, TTS, and call consumption with per-model pricing and analytics APIs.
• Implemented multi-provider LLM, STT, and TTS failover, along with a centralized event ledger for observability, reliability, and production monitoring.
• Delivered key platform features including security & compliance, audit logging, data retention, IP allowlisting, PDF compliance reports, and contributed to the real-time voice pipeline with WebRTC, Twilio, hybrid intent routing, and LLM orchestration.

Associate Software Engineer at Partex.AI (formerly Innoplexus) (2024-08 – 2026-01)

• Asset42 (Next.js, Node.js, MongoDB, GraphQL, Docker, Redux): Built and scaled a pharmaceutical drug-asset intelligence platform used by 30+ enterprise clients to visualize competitive pipeline data. Cut API latency by 35% through request debouncing, optimized REST/GraphQL query design, and WebSocket streaming for live asset updates. Designed interactive analytics dashboards with charting libraries and Auth0-secured role-based access.
• Meta D3 (Redis, Docker, ElasticSearch, React, TypeScript, Node.js, gRPC, NextJS): Implemented a Redis-backed job queue for task scheduling and priority handling, and containerized microservices with Docker for consistent multi-environment deployments. Reduced dashboard load time 50% via route-level lazy loading and code splitting; integrated Auth0 SSO for 100k+ users.
• TranslateKaro (Next.js, Auth0, LangGraph, TanStack Query, Node.js, Payment Gateway): Shipped an AI document-translation SaaS processing large documents through chunked RAG and agentic refinement flows, with Auth0 authentication and payment-gateway billing. Built a custom real-time editor with side-by-side source/translation preview and an embedded AI assistant for contextual refinement, cutting manual post-editing effort. Architected backend APIs and scalable data-fetching/caching layers with TanStack Query to keep large-document loads responsive.
• AI-ectd (Next.js, Auth0, LangGraph, TanStack Query, Node.js, Weaviate, Flask, Fast API): Built an automated CTD/eCTD generation platform converting unstructured regulatory source documents into submission-ready CTD modules, collapsing authoring time from weeks to hours. Designed a multi-agent LangGraph workflow (extraction → validation → assembly) with QA loops, citation grounding, and hallucination guardrails to keep generated sections audit-defensible. Engineered a RAG and vector-search pipeline on Weaviate with semantic chunking and re-ranking, measurably improving retrieval precision over baseline dense-only search. Owned the full Next.js frontend and backend services for document upload, job tracking, and human-in-the-loop review.
• Submission Assistant (React, Redux, MSAL, Next.js, Python, Django, SQL): Built core frontend for a compliance automation tool that reduced legal/finance documentation effort by 70%, sharply improving turnaround on government notices. Engineered a custom in-app PDF/DOCX editor for real-time document interaction, enabling teams to draft and refine submission files without leaving the platform. Developed responsive React interfaces from Figma designs and optimized Redux state flows for performance.

Project Trainee Intern at Omniscient Software Pvt. Ltd. (2024-01 – 2024-05)

• Led end-to-end development of payments and account-processing modules, improving transaction processing efficiency by 20%.
• Built a multi-format account-statement ingestion system (ICORE and other file types), generating detailed statements and reconciling 1,000+ transactions with higher accuracy.
• Extended the payments pipeline to support multiple PAIN / ISO 20022 formats, converting 1,000+ files per day into bank-specific formats for automated downstream integration.

Master of Science in Computer Science – Fergusson College (2024)

Bachelor of Science in Computer Science – Modern College of Arts, Science & Commerce (2022)