Open to Forward Deployed Engineer / AI Platform Engineer roles

Jinson EB

AI Engineer

I build and deploy production AI orchestration systems end-to-end — from LLM pipelines to observability — embedded inside a real enterprise client, working directly with their business stakeholders to ship what actually works within real constraints.

Currently at Assuretech Business Solutions

Backend Engineering AI Systems Distributed Systems Platform

View Projects Contact Download Resume

STACK / Python·OpenTelemetry· Qdrant·PostgreSQL· AWS·Docker

public/images/jinsoneb.jpeg

01 / About

Engineer focused on systems that hold up in production.

I'm a software engineer with 3 years building and deploying production AI systems embedded inside a single enterprise insurance client — Dubai Insurance (AED 3.44B annual revenue, AED 4.3B in assets). I own an AI orchestration deployment end-to-end inside their production environment, cutting a manual verification workflow from 40–60 hours per employee to 1–2 hours — a ~95% reduction.

The system runs as a Python Windows service polling their production database and routing records through a LangChain + Google GenAI orchestration pipeline, instrumented with OpenTelemetry (Arize Phoenix) for per-case cost, latency, and evaluation tracking. I act as the technical point of contact with their Claims Manager, lead a team of 5 engineers through delivery, and have turned the platform into a repeatable rollout now onboarding a second client. I'm looking to bring that same ownership to a product-based AI engineering team as a Forward Deployed Engineer or AI Platform Engineer.

Experience: 3+ yrs
Domain: Enterprise Insurance
Focus: AI Orchestration
Based in: Kochi, India

02 / Experience

Where I've worked

AI Software Engineer · Assuretech Business Solutions

01/2026 – present

Owned an AI orchestration deployment end-to-end inside an enterprise insurance client's production environment, acting as the technical point of contact and leading a team of 5 engineers through delivery.

Cut a manual verification workflow from 40–60 hours per employee to 1–2 hours (~95% reduction), shipped as a Python Windows service polling the production database and routing records through an LLM-driven pipeline
Built the agentic orchestration logic with LangChain and Google GenAI, instrumented with OpenTelemetry (Arize Phoenix) for per-case cost/latency tracking and versioned, effectiveness-scored prompts
Authored deployment and feature documentation that turned the platform into a repeatable rollout, onboarding a second client

PythonLangChain Google GenAIOpenTelemetry Arize Phoenixpywin32

Software Engineer · Assuretech Business Solutions

03/2025 – 01/2026

Built internal platforms automating client-facing business workflows.

Built an internal platform automating proposal filing and invoice collection, cutting company revenue leakage by 99%
Developed a mobile application delivering approval workflows for upper management

.NET FrameworkOracle

Junior Software Engineer · Assuretech Business Solutions

09/2023 – 03/2025

Started out on the frontend team building customer-facing support tooling.

Collaborated with senior developers to build a CRM portal for customers to raise and track grievances

TypeScriptAngular Docker

03 / Skills

Capabilities

Languages

C#Python TypeScriptJavaScript SQL

Backend

.NETASP.NET Core FastAPIREST

Frontend

AngularTypeScript TailwindHTML/CSS

AI

LLMsRAG LangChainMCP Agents

Cloud

AWS

DevOps

DockerGitHub

Databases

PostgreSQLOracle RedisNeo4j Qdrant

Observability

OpenTelemetryGrafana PrometheusPhoenix

04 / Featured Projects

Selected work

CURRENT / IN PROGRESS

AI Platform — Central Intelligence Layer

Architecting a central AI platform that acts as the brain behind fragmented, client-deployed AI systems — unifying them into one intelligence layer that gets smarter with every deployment.

Unified data layer across relational, vector, and graph stores for retrieval, similarity search, and relationship modeling
Designed to compound — every client deployment feeds data back into the platform, improving future rollouts

PythonPostgreSQL QdrantNeo4j DockerAWS

Confidential client platform — details on request

PERSONAL PROJECT

Aether — RAG Pipeline

Self-built Retrieval-Augmented Generation pipeline deployed on AWS over a corpus of database-design papers, with vector embeddings, semantic search, and an optional @web_search tool fallback.

Wired to OpenObserve and Arize Phoenix (OpenTelemetry) for full-pipeline monitoring
Containerized with Docker for publishing and deployment

PythonAWS OpenTelemetryDocker

aether-rag-pipeline.duckdns.org Source

05 / Live Products

Deployed & running

Live

Recovery Orchestration Platform

Built for a Dubai-based insurance client, integrated with their core insurance product. An LLM verifies scattered claim assets against policy ground truth, runs them through the client's business validation rules, and auto-creates the claim file — or flags a validation error for human review. Developed alongside the client's claims manager; cut per-claim processing from 7–8 business days to roughly half a day to review all ~3,000–4,000 monthly records.

LLM DecisioningInsurance Document VerificationHuman-in-the-Loop

Confidential client platform — details on request

Live

Proposal-to-Cash Automation Platform

Internal platform automating the company's full proposal-to-cash lifecycle — from client proposal drafting through order conversion, milestone invoicing, and payment reconciliation. Covers payment terms, products, clauses, work/personnel assignment, cost and profit metrics, amendment versioning, and a three-level approval workflow; converts accepted proposals into orders; runs a timed invoice-raising plan tied to milestone completion routed through accounting approval; and automates invoice delivery with partial/full payment tracking.

.NETSQL Workflow AutomationApproval Routing

Internal company platform — details on request

Live

Salesforce CRM Data Integration Pipeline

Bi-directional data pipeline syncing our core system with a client's Salesforce CRM, giving their support desk direct visibility into customer data for faster issue resolution. Exposed endpoints let the client's support team update customer details and register claims directly from Salesforce, feeding straight back into our core system.

SalesforceREST APIs Data Integration.NET

Internal company platform — details on request

Live

Complaint Registration Portal

Complaint management portal for a client, with separate internal and external customer registration facilities. Tracks every complaint through resolution with time-based auto-escalation, assigns issues to team members, and communicates with external customers over email and WhatsApp.

.NETSQL WhatsApp APIAuto-Escalation

Internal company platform — details on request

06 / AI Engineering

Building agentic systems, retrieval pipelines, and durable workflows.

AI Agents

Tool-using agents with planning, memory, and guardrails.

LLMs

Prompt design, structured output, evaluation and routing.

RAG

Retrieval pipelines with chunking, reranking, and grounding.

Temporal

Durable, fault-tolerant orchestration for long-running work.

Vector Databases

Semantic search over embeddings with Qdrant and friends.

OpenTelemetry

Tracing and observability across model and service calls.

LangChain

Composable chains and integrations for LLM applications.

MCP

Model Context Protocol servers exposing tools and data.

Workflow Automation

Event-driven automation wiring systems and humans together.

07 / Tech Stack

Tools I reach for

08 / Certifications

Credentials

Ekathon 2026

AI Impact Summit — Medical Automation CRM System

2026

Google Developer Meet

Bangalore

2025

Cloud Native Summit

Kerala

2025

09 / Education

Education

M.Sc. Mathematics (AI focus)

2025 – present

Vivekananda Global University

B.Tech, Mechanical Engineering

2017 – 2021

Cochin University of Science and Technology (CUSAT)

10 / Contact

Let's build something reliable.

Open to conversations about roles, collaborations, and interesting problems.

jinsoneb@gmail.com