Complex data.Practical software.Clear purpose.

I led architecture and implementation across the full stack. I built Python services using scikit-learn on AWS EC2 to process sensor observations through Redis Streams and classify movement patterns, storing results in PostgreSQL and Supabase. I built the authenticated dashboard, visual map and client APIs, plus email alerts and PDF reporting stored in S3. I deploy on Vercel through its GitHub integration, with Cloudflare and Turnstile, and own monitoring and analytics.

Built with
Python · scikit-learn · AWS EC2 · Redis Streams · PostgreSQL · Supabase · S3 · Next.js · TypeScript · React · Vercel · Cloudflare · Turnstile

I led full stack development and client engagement from concept to production. I built vendor invitations, one-time password access, document uploads, staged approvals, corrections and resubmissions. Application and review data live in PostgreSQL and Supabase, with encrypted documents in S3 and Resend handling email workflows. I improved security, database performance and usability after taking over the early prototype.

I own architecture and weekly planning, with regular client communication throughout delivery. I mentor a junior developer, giving them room to explore solutions and supporting their work through code review and testing.

Built with
Next.js · TypeScript · React · PostgreSQL · Supabase · S3 · Resend · APIs · Vercel

I work as a data scientist within Momentum’s Disruptive Innovation (Advanced Analytics & AI) team, a medium-sized multidisciplinary team. The engagement accounts for up to 60% of my working time, alongside other MUNUS projects. I develop, train and compare candidate models in Python using separate training, validation and test datasets. I evaluate relevant metrics and document comparisons to guide model selection, retaining existing models when alternatives do not improve performance. I assess model drift and retrain when it is detected.

I transform deeply nested credit and risk records into consistent inputs, with safeguards for missing fields, inconsistent structures and differing formats. I built and maintain five production APIs in Python using FastAPI, including two that run model predictions, and test input handling and responses against expected outputs.

I also built the Python, HTML and CSS reporting tool and collaborate with cloud engineers and back-end developers in India on AWS deployment, integrations, Datadog monitoring and ongoing production support.

Built with
Python · FastAPI · HTML · CSS · AWS EC2 · S3 · CodeCommit · Deployment pipelines · Datadog

I built BlueCrane from the ground up, beginning in 2023 and bringing it into production in my subsequent role. Teams configure room and area questions; students submit answers, photos and additional issues. I built Python workers for data requests and integrated ERIS RiseStudent through SSO and REST APIs, returning inspection data to the client. Supabase stores inspection records, S3 stores photos, and automated emails keep users informed.

Built with
Python · Next.js · TypeScript · React · Supabase · SQL · APIs · Vercel · S3 · Cloudflare

I built ingestion, cleaning, validation and transformation pipelines for large weather datasets. I implemented the processing that applies the project’s ML models to fill gaps in the data, ran it on AWS EC2 and delivered the results to the client’s S3 bucket.

Built with
Python · AWS EC2 · S3 · REST APIs · Queues

I progressed from building client dashboards to owning the platform and releasing V2. I built a client setup application and dashboards, with shared Python and SQL processing to retrieve MyBuildings data, calculate scores and store prepared results in PostgreSQL and Supabase for each client’s separate dashboard. This reduced duplicated code and improved load times. I added MQTT sensor feeds, Redis sessions, row-level security, encrypted connections and backups. A separate FastAPI service accepts client sensor data and triggers automated work requests in MyBuildings.

Built with
Python · SQL · Next.js · TypeScript · React · FastAPI · PostgreSQL · Supabase · MQTT · Redis · Vercel · DigitalOcean · Ubuntu · Cloudflare

I built the whole system and client integrations. Python consumers process tracking data from RabbitMQ, analyse patterns and publish results back through a queue. I delivered fleet event detection and heatmaps, deployed on a DigitalOcean Droplet with Nginx, and supported the system in production for a year.

Built with
Python · RabbitMQ · APIs · Redis · Plotly · React · HTML · DigitalOcean · Nginx

I built and deployed the portal from scratch. Operators review incoming issues, merge duplicates, decline reports or send accepted reports to MyBuildings as work requests. I built the application and API endpoints with Next.js and TypeScript, using PostgreSQL and Supabase to store records, plus work-request delivery and status retrieval. Updated statuses, or the stored response when no update is available, let users follow progress through GoWaterfall.

Built with
Next.js · TypeScript · React · PostgreSQL · Supabase · APIs · Vercel · Cloudflare

I built and supported the system from scratch. Python processing cleans uploaded CSV exports and calculates gross profit and stock reconciliation. Records and calculated results are stored in Supabase and displayed in authenticated Next.js and TypeScript dashboards. I owned development, delivery and support.

Built with
Python · Next.js · React · TypeScript · Vercel · Dash · Plotly · Supabase · DigitalOcean · Cloudflare