Muhammad Nebil

AI/ML engineer with hands-on experience in LLM evaluation, RLHF pipelines, and prompt engineering, backed by a full-stack development foundation in Android and PHP/MySQL systems. Evaluated 500+ LLM responses weekly at Innodata Inc., improving instruction-following accuracy through structured rubric design and bias/hallucination detection. Built and deployed 3 production Android applications and REST API backends serving 500+ concurrent users.

Seeking AI/ML Engineer or Machine Learning roles applying LLM fine-tuning, RAG pipeline development, and model evaluation at scale.

AI/LLM Analyst at Innodata Inc. (2025-12 – Present)

• Evaluated 500+ LLM-generated responses weekly using Python scripting and structured rubrics, improving evaluation throughput 40% over manual workflows and feeding higher-quality signals into RLHF fine-tuning cycles.
• Maintained 95%+ inter-annotator agreement across 10+ concurrent annotation projects by designing evaluation rubrics and running statistical QA audits before dataset submission.
• Documented 100+ model failure cases (hallucinations, reasoning gaps, demographic bias), delivering feedback adopted into the next fine-tuning sprint and improving instruction-following accuracy on benchmarks.
• Prototyped prompt-engineering templates using Claude and GPT-4 APIs to stress-test edge cases before human annotation, cutting annotation rework rate ~20% across 3 concurrent projects.

Software Developer at WinWay IT Park (2025-01 – 2025-12)

• Built and shipped 3 production Android applications (Java, MVVM), reducing screen load latency ~30% through asynchronous data pipelines and optimized LiveData binding; apps served 500+ concurrent users.
• Designed PHP/MySQL backend services (authentication, session management, RBAC) and automated deployments via Jenkins CI/CD, cutting release cycle time 20%.
• Built 10+ REST API endpoints with input validation, rate limiting, and error handling, reducing data fetch latency 25%; documented API contracts using Swagger.
• Applied SQL performance tuning (indexing, query optimization, stored procedures), cutting average query execution time 40% under 200+ concurrent database sessions.
• Deployed the ZeemPlus billing system to AWS EC2 with S3-backed file storage, establishing baseline cloud architecture and reducing dev/production parity issues.

Master of Computer Applications in MCA – University of Calicut, CCSIT Palakkad (2023 – 2025)

Bachelor of Science in Computer Science – University of Calicut, IHRD College Kottayi (2020 – 2023)