Experience

Deep-dive breakdown of systems engineered, evaluation frameworks built, and workflow validation across Dubai, Trivandrum, and London.

Isometric multi-agent workflow orchestration and automated browser system
Dubai · Jul 2025 – Jun 2026

Agentic & Django Automation

3-stage automated invoice reminders, browser-automated grant monitoring across 5+ sources, and SQL/API validation.

Isometric multi-document RAG retrieval and evaluation pipeline
Trivandrum · Mar 2025 – Jun 2025

Grounded RAG & LLM Evals

Multi-document retrieval assistant evaluated with LLM-as-a-Judge, embedding similarity, BLEU, and ROUGE in Dockerized REST services.

Enterprise NLP analytics dashboard representing clustered review sentiment analysis
London · Dec 2024 – Feb 2025

NLP Review Intelligence

57,391 cleaned software reviews analysed with topic discovery, 82.94% emotion classification baseline, and PM prioritization views.

Cross-Role Delivery & Verification Matrix

3 International Teams · Chronological Comparison
Organization & Tenure Role Focus Core Systems Shipped Verification & Impact
Wadi Al Ain
Dubai · Jul 2025 – Jun 2026
Software Developer 3-stage invoice reminder workflow; agentic Python & browser-automation grant monitor across 5+ funding sources; Django backend APIs. Targeted 40% reduction in manual follow-up effort; validated SQL integrity, input handling, failure recovery, and multi-step consistency.
Ladder7 Nextstep Solutions
Trivandrum · Mar 2025 – Jun 2025
Data Science Associate Multi-document RAG research assistant built with dense embeddings, vector search, prompt refinement, and Dockerized REST APIs. Evaluated relevance, correctness, faithfulness, and hallucinations using LLM-as-a-Judge, embedding similarity, BLEU, and ROUGE.
First Quadrant Labs
London · Dec 2024 – Feb 2025
Data Science Intern NLP review analytics across 57,391 cleaned software reviews, interactive PM evidence dashboard, and incremental CSV ingestion pipeline. 82.94% accuracy emotion classification logistic regression baseline with schema validation, duplicate detection, and import logging.

Software Developer

Wadi Al Ain Documents Clearing Services

Jul 2025 – Jun 2026 Dubai
  • Automated invoice tracking and a 3-stage payment-reminder workflow, targeting a 40% reduction in manual follow-up effort through structured payment tracking and automated reminders.
  • Developed components of an agentic grant-monitoring system using Python and browser automation to collect opportunities, track deadlines, and assess eligibility across 5+ funding sources.
  • Validated backend APIs, data-processing logic, input handling, and SQL operations using Python and Django; investigated defects and verified fixes.
  • Tested multi-step processing workflows for data consistency, failure handling, and reliable execution across integrated application components.

Systems Built

Three-stage invoice follow-up state tracker and an autonomous Python + browser automation pipeline aggregating grant criteria and deadlines across 5+ external portals.

Validation & Reliability Focus

End-to-end Django REST endpoint checks, SQL query verification, edge-case input validation, defect root-cause investigation, and multi-step failure-recovery testing.

Python · Django · Agentic AI · Browser Automation · SQL · REST APIs

Data Science Associate

Ladder7 Nextstep Solutions LLP

Mar 2025 – Jun 2025 Trivandrum
  • Developed and evaluated a multi-document RAG research assistant using LLMs, embeddings, vector search, and retrieved-context grounding.
  • Designed evaluation scenarios to assess generated responses for relevance, correctness, faithfulness, consistency, and potential hallucinations.
  • Applied prompt refinement and LLM-as-a-Judge evaluation concepts alongside embedding similarity, BLEU, and ROUGE to assess output quality.
  • Integrated and tested AI workflows through REST APIs and Dockerized services, validating request handling, response structure, and component interoperability.

RAG Architecture

Multi-document ingestion, dense embeddings, vector similarity retrieval, and strict context-grounding prompts served through containerized REST APIs.

Quality & Hallucination Evals

Evaluated relevance, faithfulness, correctness, and consistency using LLM-as-a-Judge scenarios, embedding similarity, BLEU, and ROUGE scoring.

LLMs · RAG · Embeddings · Vector Search · BLEU · ROUGE · REST APIs · Docker

Data Science Intern

First Quadrant Labs

Dec 2024 – Feb 2025 London
  • Analysed 57,391 cleaned software reviews to identify recurring customer pain points, sentiment patterns, and product feedback themes.
  • Built text-classification and topic-discovery workflows, achieving 82.94% accuracy with a logistic regression baseline for emotion classification.
  • Developed an interactive dashboard with issue-level evidence, product comparisons, and prioritization signals for product managers.
  • Implemented CSV ingestion with validation, duplicate detection, import logging, and reusable model artifacts for incremental review analysis.

ML & Topic Modeling

Processed 57,391 software reviews to surface product pain points and built a reproducible logistic regression emotion baseline hitting 82.94% accuracy.

Ingestion & PM Dashboard

Schema-validated CSV uploader with duplicate detection, import logs, reusable model artifacts, and comparative issue-level evidence for product decisions.

Python · Scikit-Learn · Pandas · NumPy · EDA · NLP Classification · Topic Discovery