Jeyasurya V · Software Engineer

I build multi-tenant platforms, enterprise integrations and production AI systems.

Go, TypeScript and Python in production — from ITSM sync engines to hybrid LLM/SLM infrastructure.

Years in production: 3+
Lower AI inference cost: ~40%
Systems in case studies: 5

Download résumé Contact

Coimbatore, India · Open to remote roles and relocation, including abroad

Fig. 00 — Five systems, one scaleSelect a site to open its case study

01 — Selected work

5 case studies

Ch. I
Enterprise AI Platform
An LLM orchestrator, a self-hosted small model and a per-tenant vector index powering summaries, RCA, search and insights.
Python · FastAPI · LlamaIndex · Milvus · llama.cpp
Fig. 01Professional · Tacbot.ai
Ch. II
Enterprise Integrations & Quality
Bi-directional ServiceNow sync without update loops, NetBox inventory, email-to-case intake and a daily end-to-end safety net.
Go · Gin · PostgreSQL · ServiceNow · NetBox
Fig. 02Professional · Tacbot.ai
Ch. III
Soulbound
A self-hosted AI companion with governed long-term memory that can explain why it believes anything.
Python · Hybrid retrieval · Embeddings · Voice · pytest
Fig. 03Personal project
Ch. IV
Cadence
An exam-preparation operating system built on spaced repetition, per-user data isolation and push reminders.
Next.js · TypeScript · Supabase · Postgres RLS · Web Push
Fig. 04Personal project
Ch. V
SpeakUp
An offline-first reading companion for Android that keeps every book's identity, position and notes safe across devices.
Flutter · Dart · Riverpod · Drift · Supabase
Fig. 05Personal project

02 — Experience

Download résumé (PDF)

07/2025 — Present
1 yr 4 mos
Tacbot.ai
Software Engineer
Coimbatore, India
Multi-tenant SaaS case-management platform for enterprise IT operations.
- Designed the LLM/SLM orchestrator that routes every AI use case between a self-hosted small model (llama.cpp on Kubernetes) and Claude on AWS Bedrock, with permission-gated fallback, rate limiting and per-call usage metering — cutting AI inference cost by about 40%.
- Built the vector search feature end to end — Milvus with per-tenant collections, structure-aware chunking of cases, alerts and inventory, and cross-encoder reranking — grounding AI search, summaries and RCA timelines in each tenant's own records.
- Built bi-directional ServiceNow sync behind a provider-agnostic integration orchestrator, with webhook echo prevention, a retry-able failure queue, duplicate handling, CMDB import and per-tenant generated Update Sets.
- Built the NetBox integration — live schema discovery, field mapping, approval-staged scheduled polling and event-rule push, reaching private networks through a tenant-selected gateway tunnel.
- Worked on the multi-tenant core — tenant-aware middleware and per-tenant database routing — and email-to-case over IMAP and Microsoft Graph with historical mailbox onboarding.
- Engineered a 53-spec Playwright suite that scaffolds its own tenants and runs business-critical vendor workflows daily in GitHub Actions; deployed OpenLIT LLM observability across environments.
Go · Gin · PostgreSQL · Python · FastAPI · Milvus · llama.cpp · AWS Bedrock · Next.js · Playwright
Case studies: Fig. 01 Enterprise AI Platform · Fig. 02 Enterprise Integrations & Quality
06/2023 — 06/2025
2 yrs 1 mo
Clar Technologies
1. Software Development Engineer06/2024 — 06/2025 · 1 yr 1 mo
2. Software Engineer06/2023 — 05/2024 · 1 yr
Coimbatore, India
Blockchain supply-chain finance and AI-assisted data tooling.
- Engineered full-stack blockchain solutions for supply-chain finance, including Digital Ledger Payment Commitment (DLPC) systems for tokenised transactions.
- Implemented mnemonic isolation with AWS Lambda and Secrets Manager to protect keys and sign transactions reliably.
- Developed an SDK and Node CLI connecting backend systems to the blockchain layer, improving maintainability by about 60%.
- Built Jenkins and Docker CI/CD pipelines that cut deployment time by about 40%.
- Built an AI-assisted annotation platform (React, OpenCV, Django REST) with dynamic schema mapping, YOLO pre-annotation and COCO, YOLO and Pascal VOC export.
Node.js · React · Django · AWS Lambda · Secrets Manager · Docker · Jenkins · OpenCV

03 — Skills

Each domain cites its evidence

Skills by domain, with the case studies and roles where they were used
No.	Domain	Technologies	Evidence
01	Languages	Go · TypeScript · JavaScript · Python · Dart	Fig. 01 Fig. 02 Fig. 03 Fig. 04 Fig. 05 Tacbot.ai Clar Technologies
02	Frontend	React · Next.js · Redux · Tailwind CSS · shadcn/ui · Ant Design	Fig. 04 Tacbot.ai Clar Technologies
03	Mobile	Flutter · Riverpod · Drift (SQLite)	Fig. 05
04	Backend	Go (Gin) · FastAPI · Node.js · Express · Django · REST	Fig. 01 Fig. 02 Tacbot.ai Clar Technologies
05	Data	PostgreSQL · Supabase · Milvus · SQLite · MongoDB · Redis	Fig. 02 Fig. 01 Fig. 04 Fig. 05 Tacbot.ai
06	AI systems	AWS Bedrock · LlamaIndex · SLM serving (llama.cpp) · Vector search & reranking · RAG · MCP · OpenLIT observability	Fig. 01 Fig. 03 Fig. 04 Tacbot.ai
07	Cloud & DevOps	AWS (S3, Lambda, Secrets Manager, IAM, API Gateway) · GCP · Docker · Kubernetes · GitHub Actions · Jenkins	Fig. 01 Fig. 02 Tacbot.ai Clar Technologies
08	Testing	Playwright · Jest · pytest	Fig. 02 Fig. 03 Fig. 04 Tacbot.ai
09	Integrations	ServiceNow · NetBox · Microsoft Graph · IMAP	Fig. 02 Tacbot.ai

04 — Contact

Let's build something that holds up.

Open to remote roles and relocation, including abroad

jeyasuryavelmurugan09@gmail.com

I build multi-tenant platforms, enterprise integrations and production AI systems.

Enterprise AI Platform

Enterprise Integrations & Quality

Soulbound

Cadence

SpeakUp

Tacbot.ai

Clar Technologies