Rohit
Jain

1. itrfa.in

   Live

   Solo project  ·  1,200+ filings, 480+ paid, first season

   Python FastAPI GCP Cloud Run Claude & Gemini Firestore GitHub Actions

   A tax-filing tool for Indian residents who hold foreign equity from RSUs and ESPPs. It reads a broker's raw export and produces the schedules and computations the return needs.

   • Statement parsers for eight brokers (Fidelity, Schwab, E*TRADE, Interactive Brokers, Morgan Stanley, Shareworks, Vested, NetBenefits), normalising different CSV, PDF, and XLSX layouts into one lot-level model.
   • Python and FastAPI on GCP Cloud Run, with Firestore, Cloud Storage, and GitHub Actions. Claude and Gemini handle extraction from broker PDFs that ship no machine-readable format.
   • Everything from design to on-call is mine, including the currency, corporate-action, and cost-basis edge cases that make this domain unpleasant.

   Open a US broker's CSV in Excel on an Indian laptop and it swaps day and month, but only on dates where both readings are valid.

   Price feeds rewrite history after a stock split. Broker exports don't. Multiply one by the other and a 5-for-1 split values a holding at a fifth.

   Ask a quote API for "TOAST INC CL A" and the only hit is a sparse Milan listing. US listings now win the tiebreak.

   1. Broker exportsCSV · PDF · XLSX Eight brokers: Schwab, Fidelity, IBKR, E*TRADE, Morgan Stanley and more
   2. Parse & extractFastAPI · Cloud Run Eight layout normalisers; Claude and Gemini for statement PDFs
   3. ReconcileFirestore · Python Lot-level model, SBI TT buying rates, splits, transfers, corporate actions
   4. OutputsITR-ready Schedule FA, Schedule CG, audit sheets and notes

2. Knowledge Atlas

   ServiceNow  ·  design owner, presented at architecture review board

   Enterprise GenAI Content Pipeline Architecture Review Board Java

   A central repository that consolidates help articles from three separate content sources, with a review workflow where generative AI identifies issues in published articles and routes them to the writers who own them.

   1. Sources Help articles from three separate content sources
   2. Consolidate One central repository for every article
   3. Review Generative AI identifies issues in published articles
   4. Route Each issue goes to the writer who owns the article

3. Retrieval-augmented search for Now Support

   ServiceNow  ·  design owner

   RAG Vector Search FAISS Embeddings LLM Grounding

   Search over support content that combines embeddings-based retrieval with an LLM to return a grounded answer, rather than a list of articles the reader still has to dig through.

   1. Corpus ingestionScheduled delta Knowledge articles and product documentation
   2. IndexingFAISS Dense embeddings with product and service metadata
   3. RetrievalTop-k Similarity plus product filters, then prompt assembly
   4. GenerationGrounded An answer grounded in the retrieved text, with source citations

More at ServiceNow

• Course recommender for Now Learning

  Latest project

  A recommendation system on Databricks that suggests the courses on Now Learning each learner is most likely to need next.

  Databricks Recommendations Now Learning

• Content tagging service

  Design owner

  Tags every article with the products and services it covers, using embeddings and FAISS for matching, so downstream systems can filter and route content without a human keeping a taxonomy up to date.

  FAISS Embeddings Automated Taxonomy Python

• AI feedback management & triage platform

  Design owner

  Designed and built the feedback management platform (collection, storage, routing, and analytics, with data retention and access control designed in), and replaced manual triage with an AI agent that classifies and routes incoming feedback directly to owning teams.

  AI Agents Triage Automation Data Retention Access Control

• Localization as a Service

  Design owner

  A self-serve translation pipeline for internal applications, replacing a manual vendor handoff and dramatically shortening release cycle times.

  Platform Services Self-serve Pipeline Internal APIs

• Best Practices Library backend

  Led a team of five

  Owned the service design, API surface, and database schema, with Cloudflare caching in front and delivery on AKS. Ran the team from requirements through to monitoring.

  Team Leadership AKS / Kubernetes Cloudflare Java & Spring Boot

• Generative AI in the internal Code Scanner

  Extended the static-analysis tool so it proposes and applies fixes for its findings instead of only reporting them.

  LLM Remediation SAST Adoption Developer Productivity

Earlier

• Alt Builder

  PeopleStrong  ·  led a team of five

  A configurable application builder that stands up client-specific HR applications with no per-client code, on Kubernetes.

  Kubernetes Angular

• Hotel discounting engine

  MakeMyTrip

  Rewritten as a rule-driven campaign system so marketing could launch promotions without an engineering release.

  Rule Engines

• Equity trade lifecycle application

  Sapient Global Markets

  Trade capture through to settlement, for global investment-banking clients.

  Capital Markets

Languages

Java, Python, JavaScript, SQL, Go (working knowledge)

Backend

Spring Boot, FastAPI, Kafka, REST, microservices, distributed systems

Data

PostgreSQL, MySQL, Redis, Firestore, FAISS, Databricks, Snowflake

Cloud & infra

Kubernetes (AKS), Docker, GCP, Azure, AWS (Bedrock, Lambda), Cloudflare, GitHub Actions

Applied AI

LLM application work with Claude and Gemini, AI agents, retrieval-augmented generation, embeddings and vector search, document extraction, evaluation harnesses Prompt caching never hit in production: the cached prefix was shorter than the model's minimum. Moving one breakpoint fixed it.

Security

Threat modelling, SAST, secure design review; security champion for my team's services

1. 2020 – now ServiceNow Staff Software Engineer (2022–), Senior Application Developer (2020–22). Hyderabad.
2. 2016 – 2020 PeopleStrong Technology Lead (2019–20), Senior Software Engineer (2016–19). Gurugram.
3. 2014 – 2016 MakeMyTrip Senior Software Engineer. Gurugram.
4. 2012 – 2014 Sapient Global Markets Associate Technology L1. Bangalore.

• B.E. in Computer Engineering, Delhi College of Engineering (now Delhi Technological University), 2008–2012.
• TOGAF 9, Level 1 and Level 2.
• Artificial Intelligence and Machine Learning for Industry, IIT Delhi, 2024.
• ServiceNow Certified System Administrator and Certified Application Developer.