spbuilds.in available get in touch

résumé

Someshwar Patil

Applied AI Engineer · Senior Full-Stack Engineer

Pune, India · open to roles · someshwar@spbuilds.in

Summary

Engineer with ~6 years building production backend services, high-volume data-processing pipelines, and payments/billing infrastructure on Google Cloud — now focused on applied Generative AI: LLM applications, RAG systems, MCP servers, AI agents, and LLM evaluation harnesses, built and deployed end to end. Fluent in Python (FastAPI, Flask) and TypeScript; I own systems from API design through deployment and production support.

Core skills

Generative AI

  • LLM apps
  • RAG
  • Vector DBs & embeddings
  • AI agents & multi-agent orchestration
  • Fine-tuning
  • LLM eval & benchmarking
  • Prompt engineering
  • Tool / function calling

AI frameworks & platforms

  • LangChain
  • LangGraph
  • MCP servers
  • Claude Agent SDK
  • Vertex AI (Gemini)
  • Anthropic Claude

Languages

  • Python
  • TypeScript
  • JavaScript
  • SQL

Backend & APIs

  • FastAPI
  • Flask
  • REST API design
  • Microservices
  • Real-time webhook pipelines
  • GraphQL
  • Node.js

Data processing

  • High-volume ingestion & sync
  • Batch processing
  • PySpark
  • Firestore
  • Storage / file pipelines

Payments

  • Stripe billing
  • Subscriptions
  • Invoicing
  • Payouts
  • Full payment lifecycle

Cloud & DevOps

  • GCP (Cloud Run, Functions, Firestore, Firebase Auth, Storage, Pub/Sub, Scheduler, Secret Manager)
  • Docker
  • CI/CD
  • Git

Frontend

  • React
  • Next.js
  • Tailwind CSS

Experience

Senior Full-Stack Engineer / Applied AI Engineer

A US-based print-on-demand SaaS · Remote

Aug 2020 – Jul 2026
  • Architected and owned the billing and subscription system on Stripe — pricing tiers, invoicing, payouts, and the full payment lifecycle for platform customers.
  • Built real-time webhook and data-processing pipelines across Shopify, Etsy, WooCommerce, and Squarespace for automated, high-volume order and inventory synchronization.
  • Owned the entire Google Cloud backend — Cloud Functions, Firestore, Firebase Auth, Cloud Storage — plus deployment and CI/CD pipelines.
  • Developed barcode-driven production workflows with automated batch processing and pick-list generation for high-throughput operations.
  • Integrated shipping platforms (EasyPost, ShipStation) and supplier APIs for end-to-end order fulfillment.
  • Designed and optimized a high-volume file-storage system for print assets, improving upload reliability and retrieval performance.

Selected projects

see all work →
  • Brook

    Current product · live

    Multi-tenant SaaS — an LLM coding agent for engineering teams in Slack. Natural-language requests route to isolated per-tenant agent sessions (Claude Agent SDK on Vertex AI) with human approval gates and file-lock concurrency guards.

  • Parchi

    Client product · live

    Multi-tenant SaaS — B2B ordering + GST billing back office for goods distributors (Next.js 15 + Firebase). A GST core built as pure, unit-tested functions: integer-paise money, CGST/SGST/IGST place-of-supply split, per-FY invoice numbering, FEFO batch allocation, append-only stock ledgers.

  • sp-evals / prompt-forge

    Tooling

    An LLM evaluation harness plus a capability-transfer pipeline: a strong judge model forges per-task prompt packs that let a smaller, cheaper model reproduce high-quality output. Four task classes frozen and benchmarked against baselines.

  • claude-vm-mcp

    Tooling

    A production MCP server (7 tools, Streamable HTTP) with a Cloud Run wake-proxy for secure, on-demand access to GCP VMs — bearer auth and audit logging.

  • ai-wire

    Live

    An autonomous AI-news publishing pipeline: multi-source ingest → two-pass Vertex AI (Gemini) curation → automated carousel rendering → human approval gate → scheduled publish.

Education

Birla Institute of Technology, Pilani

Computer Science