Godson Ajodo

Agent infrastructure — capability delegation, agent runtimes, deterministic verification

I build infrastructure that makes AI agents safe to give real authority to: capabilities that narrow but never widen across a delegation tree, runtimes where permissions and memory are composable at runtime, and enforcement that stays deterministic instead of asking another model. Before that, two summers as a data engineer at Microsoft and the wireless stack for robot teleoperation at Efference AI.

Selected work

private
code
code
code
private

Archive

  • AgentMkt

    Lightning-native marketplace routing paid work to agents or humans, with escrow.

  • Mnemosyne

    Local-first terminal context capture, queryable in natural language, no network.

  • BAA World-Builder

    HUD v6 RL environment where an agent builds and plays browser games, GRPO on Modal.

  • ARC-AGI-3 Multi-Hypothesis Agent

    World-model agent that keeps competing hypotheses alive instead of committing early.

  • MapReduce Framework

    Google's MapReduce paper implemented as a distributed gRPC framework. writeup →

  • LabPilot

    Wireless certification testing driven by 13 Gemini agents, webcam photo to report.

  • Steganography

    Rust CLI hiding AES-256-GCM payloads in the DCT frequency domain of images.

  • AutoSec

    Multi-agent pipeline that scans, triages, and patches vulnerabilities via PRs.

  • MelodyTransfer

    Transformer symbolic music style transfer with rhythm-constrained sampling.

  • TouchGrass

    Social app gamifying real-world hangouts, scoring covered by property-based tests.

  • Automata-Diags

    Open-source automata theory toolkit with instant visualization. writeup →

  • MediBot

    Symptom description helper for non-native English speakers.

  • Back to the Present

    Climate simulation with LLM storytelling — People's Choice, NASA Space Apps 2024.

  • Chess Engine

    Minimax with alpha-beta pruning and Zobrist-hashed transposition tables.

Writing

Notes and writeupsAgent infrastructure, distributed systems, and what I have been building.3 pieces →

Experience

Member of Technical Staff at Efference AI
August 2025 - March 2026
  • ›Owned the wireless communication stack for cross-country robot teleoperation: Bluetooth/Wi-Fi packet protocols, transport to the control APIs, and model integration.
  • ›Helped build a video database for robotics, world models, and augmented reality.
Data Engineer Intern at Microsoft
May 2025 - August 2025
  • ›Expanded and scaled the assignment verification tool for the Xbox Gaming Experimentation team.
  • ›Reduced scorecard generation and assignment validation time from 1 day to under 10 minutes.
  • ›Designed and deployed nine management-group-specific NRT snapshot tables with 28-day retention.
Data Engineer Intern at Microsoft
May 2024 - August 2024
  • ›Engineered automated data pipelines for recurring reporting workloads, replacing manual extract-and-load steps.
  • ›Created and maintained 10+ automation scripts covering routine data engineering workflows.
  • ›Partnered with the Gaming for Sustainability Engineering team to deliver three curated datasets.
Research and Development Partner at VibeMap
September 2022 – April 2023
  • ›Researched and consulted with application developers to propose gamification features.
  • ›Used Javascript and ML algorithms like NLP and sentiment analysis to create suggestions.

Stack

Languages

Python · Rust · C/C++ · TypeScript · SQL · SystemVerilog · KQL

Frameworks

PyTorch · Transformers · Hugging Face · vLLM · LangGraph · FastAPI · Pydantic · Next.js · React · Tailwind CSS

Infrastructure

Modal · Docker · Kubernetes · Terraform · Azure · AWS · PostgreSQL · SQLite · Redis · Kafka · Apache Spark · Airflow · gRPC · MCP

Education

Minerva University

Bachelor of Science in Computational Sciences

Graduated May 2026

Coursework

Theory of Computation·Optimization