Mohammed Nihad | Portfolio

> INITIALIZING ROLE...

> ESTABLISHING SECURE CONNECTION...

0%█

[ OK ] SYSTEM AUTHENTICATED

Buildify

[ ACTIVE ]

Software Engineering Intern

December 2025 - Present

+30% returning users-30% setup timeHuman-reviewed AI

▸As the sole fullstack engineer on this project, built Retention Autopilot, an agentic AI system designed to reduce churn for the small businesses on the platform.
▸On every customer login, the system pulled usage data, segmented users automatically, and generated targeted campaigns, renewal discounts, feature nudges, ROI summaries, upgrade prompts, or re-engagement tips based on journey stage.
▸Each campaign shipped with an A/B-tested variant and human approval before launch, driving a 30% improvement in 30-day returning users and a 30% reduction in campaign setup time.

STACK://

Agentic AIUser SegmentationA/B TestingRetention Campaigns

> INITIALIZING ROLE...

0%█

[ OK ] SYSTEM AUTHENTICATED

Bloom Energy

[ INCOMING ]

Software Engineering Intern (Incoming)

June 2026 - August 2026

30,000+ systemsReal-time monitoringIncoming role

▸Unified millions of records across Salesforce, OSI PI, AWS Redshift, and internal systems into an internal diagnostic tool, enabling RMCC engineers to look up active escalations, match historical fault patterns, and receive root cause analysis with fix recommendations, cutting troubleshooting time by 80% across 30,000 monitored fuel cells.
▸Built and shipped a production visual workflow builder using React Flow, FastAPI, and MongoDB, enabling engineers to assemble reusable node-based DAGs and reducing hand-authoring time for new automation scripts by ~80%.
▸Architected a versioned compiler pipeline that translated DAGs into validated, Ruff-linted notebooks, configs, and pipeline artifacts, incorporating cycle detection and lazy-cached introspection; adopted across a team of 12.
▸Optimized AWS Redshift queries by refactoring nested query patterns and simplifying data-access logic, reducing query runtime from ~50s to <10s; then built an MCP server to expose the data to engineers in local dev environments.

STACK://

KubernetesGrafanaFastAPIBackend Tooling

> INITIALIZING ROLE...

0%█

[ OK ] SYSTEM AUTHENTICATED

Research Paper

[ RESEARCH ]

Co-Author / Research Contributor

June 2025 - December 2025

1.1B vs 8B modelsFeedback distillationBaseline benchmarking

▸Co-authored an unpublished research paper introducing a feedback distillation framework for improving LLM reasoning. The system pairs a large expert model with a lightweight student model (TinyLlama 1.1B vs. LLaMA-3 8B), where the student learns to match expert-quality feedback in real time through knowledge distillation, eliminating the need for a separate fixed amateur model.
▸Contributed to both the implementation and experimental evaluation, benchmarking the framework against CLEAR, Chain-of-Thought, and Chain-of-Draft baselines across math reasoning, instruction following, and reading comprehension tasks.

STACK://

LLMsKnowledge DistillationEvaluationReasoning

> INITIALIZING ROLE...

0%█

[ OK ] SYSTEM AUTHENTICATED

STEMKasa

[ COMPLETED ]

Software Engineering Intern

June 2025 - August 2025

Led 3 internsGitHub Actions CI/CD+25% engagement

▸Led 3 interns to rebuild a 17-file monolith as a MERN app, implementing component-based design, REST APIs, and integration tests to preserve 100% functionality.
▸Architected and deployed a full CI/CD pipeline using GitHub Actions, automating build, test, and deployment workflows to eliminate manual deployment errors across staging and production environments.
▸Engineered an AI tutoring layer with GPT-powered adaptive flashcards and lessons, then integrated secure auth and Stripe subscriptions, driving a 25% engagement increase across 3 schools.

STACK://

MERNGitHub ActionsStripeGPT Tutoring

> INITIALIZING ROLE...

0%█

[ OK ] SYSTEM AUTHENTICATED

Universacare

[ COMPLETED ]

Software Automation Intern

June 2023 - September 2023

+11% productivity15 reusable components2+ hours saved/submission

▸Built a Python Selenium web scraping workflow to extract daily Home Health Aide data, automating manual work and improving productivity by 11%.
▸Developed 15 reusable React components and integrated a role-aware online application form into the production website, improving UI consistency/access control and saving 2+ hours per submission.

STACK://

PythonSeleniumReactAutomation