Mapping Infra

Developed end-to-end Snowflake DE/ML demo leveraging Streamlit Dashboarding, Cortex Analyst/ML and FactSet RBICS data, enabling real-time market segmentation and competitive benchmarking for over 40,000 companies.

• Scripted logging behavior and memory leak mitigation for 15+ Sensu and InfluxDB relay nodes, scheduled on cronjobs.

• Built metrics analysis engine for 100+ Google Kubernetes Engine (GKE) nodes with Python to recommend certain pods downscale, freeing over 200GB of overallocated memory/disk to optimize automatic node scheduling capabilities.

• Created and deployed a set of middleware to filter corrupted InfluxDB logs on 10+ relay nodes, implementing concurrency control to lock and ensure file integrity.

• Developed a GenAI tool leveraging a T5 transformer model hosted on a GCP Vertex AI endpoint that integrated with MongoDB’s vector search feature, a custom Flask backend and a React frontend to provide a chatbot service to efficiently query information from large documentation hubs. Placed top 10 in global company hackathon.

• Utilized custom Data Distribution Software (DDS) and Pub-Sub asynchronous architecture to build custom InfluxDB query generation & processing library to streamline client DB access and support 30+ internal microservices.

• Automated reconstruction and serialization of gRPC and Protobuf messages from raw Flux data to relay library results, reducing message size by 70% and increasing transmission speed by 5x.

• Developed a high-performance, low-latency genetic algorithm framework with C++ tailored to optimize the plan generation, coordination and pathfinding capability of a heterogenous fleet of 10+ unmanned aircraft.

• Built custom heuristics for path generation and designed a factory architecture for individual task construction, improving pathfinding speed by over 1000% for 10-aircraft missions compared to uninformed algorithms.

• Spearheaded development of an experimental framework to validate efficiency of a novel facial coding software and derive actionable insights prior to B2B SaaS model launch.

• Recruited and managed of a diverse cohort of 150+ respondents, ensuring successful collection, cleansing and processing of 11,000+ data points.

• Leveraged Pandas, Seaborn and Matplotlib Python modules to extract and visualize metrics from raw model feedback for high-impact client presentations with President & VP.