Resume
Applied AI Engineer | Forward-Deployed Full-Stack Engineer
Applied AI and forward-deployed engineer with customer-facing delivery experience. Works directly with operators and domain experts to turn ambiguous workflows into production systems across web, mobile, backend, AI agents, and cloud infrastructure.
Education
Michigan State University, College of Engineering Bachelor of Science in Computer Science, 2021–2024 GPA: 3.84 Minor: Computational Mathematics, Science & Engineering Business Cognate
Professional Experience
Forward-Deployed Full-Stack Engineer
Airlink Flight School · 2025–Present
- Led technical discovery directly with the flight-school owner, instructors/pilots, and administrators, mapping how they used FlightCircle alongside paper records, manual scanning, file storage, and email-based FAA submissions.
- Owned end-to-end delivery of a web and mobile operations platform covering admissions, scheduling, grading, lesson planning, and reporting for approximately 40–50 recurring users across multiple roles.
- Digitized the recurring student-grading process, eliminating paper grade sheets, manual scanning, record lookup, and email-based submissions; reduced administrative workload by approximately 60–70%.
- Designed an offline-first grading workflow for instructors operating in-flight on unreliable connections, using local persistence and conflict-aware synchronization to preserve training records across devices.
- Built the initial platform with Python, FastAPI, Firebase, and GCP, then re-architected and migrated the core platform to TanStack Start, PostgreSQL, Better Auth, Neon, Railway, and Cloudflare as its data, authentication, deployment, and security requirements matured.
- Owned production delivery across application architecture, data migration, authentication, deployment, Cloudflare WAF, and Turnstile.
Founder & Applied AI Engineer
Allwhile · 2026–Present
- Conducted discovery with 10+ independent restaurant owners, including owners operating without intelligence tools because of cost and time constraints, and owners dissatisfied with narrow, single-platform insights.
- Designed and built Allwhile as an agentic “invisible teammate” that automates competitor discovery and monitoring, business-performance and financial tracking, campaign follow-up, and local opportunity detection around events and demand signals.
- Built Zen mode, a low-overhead conversational interface that lets owners ask Allwhile for business and market intelligence without navigating traditional dashboards.
- Translate recurring operator pain points into product workflows, data models, agent behavior, reviewable actions, and full-stack product iterations.
Software Engineer
One Community Global · 2024–2025
- Built and maintained full-stack features for the open-source Highest Good Network using React, Node.js, and MongoDB; fixed role-based permission leaks across the application.
- Improved frontend performance by 20–30% through caching, redundant API-call removal, and code optimization; reviewed 40+ pull requests, authored 15+, and helped coordinate a seven-person engineering team.
Info-Tech Assistant Intern
Computational Mathematics, Science & Engineering, Michigan State University · 2023–2024
- Automated departmental data-management workflows using Python, C++, SQLite, ORCID, and Zotero APIs, replacing manual spreadsheet-based processes.
- Built publication tracking and reporting tools that consolidated research, funding, and publication data for faculty and department administrators.
- Migrated outdated data structures into a more accessible format, improving how department staff retrieved and maintained information.
Selected Projects
See All Projects → for a complete list.
OnlyAgents · AI Agent Runtime
Go, WebSockets, SQLite, LLMs, Event Bus, Tool Calling
- Built an open-source, model-agnostic runtime for autonomous AI agents with executive-worker orchestration, event-driven coordination, tool calling, connectors, and persistent memory.
- Designed a kernel-mediated execution model with a typed event bus, preventing direct agent-to-agent calls and centralizing lifecycle management, routing, and capability boundaries.
- Implemented dynamic tool gating and on-demand tool groups so agents load only the capabilities relevant to the current task, reducing unnecessary context overhead and tightening permission boundaries.
- Built layered memory across working context, semantic episode retrieval, time-scoped knowledge-graph relationships, and behavioral pattern extraction.
- Optimized the runtime into a 36 MB single binary with approximately 27 MB idle memory and 31 ms startup time.
Pickle · Private Remote Development Lab
Go, tmux, WebSockets, Tailscale, Docker, PWA
- Built a Go-based remote development environment exposing persistent tmux sessions, shells, local projects, development servers, and Docker Compose services through a browser and mobile-friendly interface.
- Added private remote access through Tailscale Serve and a mobile home-screen experience while keeping development sessions and project data on the host.
- Added process-aware support for Codex, Claude Code, OpenCode, Pi, and Hermes, with opt-in notifications for completed work and required approvals.
Text-to-Action · Natural-Language Action System
Python, Transformers, PyTorch, Semantic Search, NER
- Built an open-source system that translates natural-language requests into executable application actions across APIs, chatbots, and voice-driven workflows.
- Implemented semantic retrieval and parameter-extraction pipelines using embeddings, LLMs, and named-entity recognition; reached 1,500+ installs.
Enhanced Video Assistant · TechSmith Capstone
Python, FastAPI, React, Docker, Azure, FFmpeg, Machine Learning
- Built and deployed an award-winning full-stack video assistant combining transcript analysis, metadata extraction, audio processing, and visual scene understanding.
- Integrated open-source machine-learning models to automate audio enhancement, generate multi-layered feedback, and identify key moments in video.
Picko · Personalized Recommendation System
Python, FastAPI, SQLModel, PostgreSQL, React Native, Semantic Search
- Built and deployed a recommendation engine using TF-IDF, cosine similarity, and semantic search for personalized movie, television, and book recommendations.
- Designed a 300,000+ item relational dataset with high-concurrency query paths and automated ingestion and refresh pipelines.
Skills
Languages: Python, Go, C/C++, TypeScript/JavaScript, SQL
AI/ML: LLM applications, OpenAI API, tool calling, agent orchestration, semantic retrieval, embeddings, persistent memory, PyTorch, scikit-learn, NLP, computer vision, recommendation systems
Backend & Data: FastAPI, Node.js, TanStack Start, REST APIs, OAuth, Better Auth, PostgreSQL, Firestore, MongoDB, SQLite
Cloud & Deployment: Docker, GCP, Firebase, Neon, Railway, Cloudflare WAF/Turnstile, Azure, Supabase, CI/CD, Tailscale
Frontend & Mobile: React, React Native, Expo, Tailwind CSS
Awards
- Auto-Owners Exposition Award — Enhanced Video Assistant, 2024
- Dean’s Scholarship — $75,000 merit-based scholarship, 2021–2024
- Best Emerging Technology Award — Brainwaive, 2023
Sriram Seelamneni • United States