heyanuroop.dev

HeyAnuroop

Software engineer. Multi-agent systems, data pipelines, and the occasional pixel worth fussing over.

hire@mail.heyanuroop.dev

ask about me

more

what I work with

Every line has a receipt

AI & Multi-Agent

05

Orchestration, retrieval, and inference — cloud and on-device.

LangGraph
Multi-agent orchestration for SHADO's cloud-first agent layer
Ollama / Local LLM Inference
Moved off LangChain in favour of direct API calls and custom orchestration
RAG Pipelines
Production retrieval systems (ChromaDB, TaxBot) and Atlas Vector Search with Qwen embeddings
llama.cpp / On-Device
Kotlin → JNI → llama.cpp pipeline for offline fallback in SHADO
Agent Architecture
Designed Haive, a three-layer framework — orchestrator, domain queens, worker agents

Frontend

04

Component architecture and data visualisation.

Next.js
Default framework for web apps; built RAISE Now and PIM's frontend on it
React / TypeScript
Component architecture across CHAOS suite apps (PIM, OWL, Aghilmort)
D3.js
Data visualisations at Gramener and Bizacuity
Redux
State management for multi-view apps with complex UI state

Backend & APIs

04

Services that stay up under real traffic.

Node.js / Express
Built leads_comms, a multi-channel dispatch microservice handling WhatsApp/SMS/email at scale
FastAPI
REST APIs for data and ML-adjacent services
Python
AI-driven analytics, embedding pipelines, and data processing
Event-Driven Architecture
SQS-based rate-limited dispatch, EventBridge-triggered automation

Cloud & Infra

04

Deployment targets, pipelines, and the glue between them.

GCP Cloud Run
Core deployment target for CHAOS suite, custom email infra, and embedding workers
AWS
Lambda, SQS, EventBridge, S3, Athena — backup automation, analytics, rate-limited messaging
Airflow
Orchestrates spend analytics DAGs feeding Parquet/S3/Athena/Metabase
CI/CD
Automated build and deploy pipelines across services

Data & Databases

04

Storage, caching, and structural analysis.

MongoDB / Atlas
Primary database across nearly every project; Atlas Vector Search for embeddings
Redis
Caching and TTL-based reply attribution in messaging systems
LanceDB
Vector store for local code-analysis tooling (GRACE)
Tree-sitter / AST
Structural code analysis and dependency graph generation

Background

Lead Software Engineer with 9+ years architecting scalable, multi-tenant SaaS platforms and AI-driven systems. Full-stack development, cloud infrastructure, and mentoring high-performing engineering teams to deliver production-grade solutions.

Years experience
9+Years experience
Products delivered
8+Products delivered
Team members led
5Team members led

Experience

  1. January 2026 — PresentCurrent

    ASBL

    Software Engineer L3 (Lead)

    Owns the lead-nurturing and data-pipeline charter end to end, leading 2 senior engineers. Built GRACE, an autonomous multi-agent dev loop, and leads_comms, a Node.js microservice that cut multi-channel response time from ~90s to under 5 seconds. Architected an Airflow spends pipeline that dropped growth-team decision latency from 24 hours to 3.

  2. March 2025 — December 2025

    Thinkhat.ai

    Lead Software Engineer

    Built a medical case-generation pipeline on DSPy, real-time AI patient voice simulation with Pipecat, and a RAG tutor grounded in lecturer PDFs. Scaled microservice success rates from 44% to 95%.

  3. April 2022 — March 2025

    Microsoft

    SDE 1

    Cut React load times by 50–60%, resolved 500+ critical S360 issues across Azure and Office 365, and automated Azure SDK upgrades — replacing a recurring day-long manual task.

  4. November 2020 — April 2022

    Woundtech

    Fullstack Engineer

    Owned features end to end in TypeScript, React and Node. Improved FHIR GraphQL performance by 70% with batching, caching and dataloader patterns — 3x throughput.

Techforce.ai · Bizacuity · Gramener — 2017 to 2020

hire me →
Anuroop Pendela

Interests

  • Reading
  • Mythology
  • Exploring Tech
  • Family Time

Beyond the Code

I believe great technology should feel simple. My approach is to start with the user's problem and work backward, cutting through complexity to find the most direct and maintainable solution. I'm driven by impact — whether that's shaving seconds off a load time or designing a system that scales effortlessly for the next million users.

What excites me is the intersection of AI and practical application: taking cutting-edge research and turning it into a robust, scalable feature that solves a real problem. My current focus is agentic AI — designing the systems that orchestrate, supervise, and ensure the reliability of autonomous agents, turning conceptual autonomy into production-ready reality.

I thrive in fast-paced, agile teams that value curiosity and data-driven decisions. When I'm not architecting systems, I'm usually at the gym, growing a tray of microgreens I'll probably forget to water, or a few episodes deep into whatever anime has my attention this month — all different ways of practising patience with a system, whether it's code, a plant, or myself.