Agentic AI Developer
Victor Ogundele
I am an Agentic AI and Automation Developer based in Kwara, Nigeria. I specialize in building intelligent autonomous systems that turn messy operational workflows into clean, reliable, automated pipelines. My expertise spans LLM-based reasoning, API orchestration, vector databases, and cloud deployment. I work effectively across remote teams and can operate fully with EST schedules.
About Me
I am an Agentic AI and Automation Developer based in Kwara, Nigeria. I specialize in building intelligent autonomous systems that turn messy operational workflows into clean, reliable, automated pipelines. My expertise spans LLM-based reasoning, API orchestration, vector databases, and cloud deployment. I work effectively across remote teams and can operate fully with EST schedules. I enjoy building systems that are not just "smart" but stable, debuggable, and production-ready. My goal is to create agents that learn with RLHF, adapt, and generate measurable impact.
Technical Skills
Languages & Core Tools
- ▹Python (advanced; backend logic, data pipelines)
- ▹FastAPI (API design and service orchestration)
- ▹Node.js (familiarity)
AI / LLM Ecosystem
- ▹Agent frameworks: LangChain, MCP, Langgraph
- ▹Embeddings & vector search: OpenAIEmbeddings, langchain-chroma, Chroma vector DB
- ▹Prompt engineering, tool calling, structured planning
- ▹Synthetic data generation & style-based fine-tuning
Backend, DevOps & Cloud
- ▹MongoDB (Odoo ERP integration; pymongo singleton architecture)
- ▹pydantic-settings for configuration
- ▹AWS / GCP deployment foundations
- ▹Logging, telemetry, observability
Additional Skills
- ▹Debugging complex distributed issues
- ▹Building feedback loops & performance evaluators for agents
- ▹Designing readable, maintainable architectures
- ▹Building observability into agents with langfuse
Featured Projects
Integrating VAPI with N8N Workflow
Here are recorded walkthroughs of my work integrating VAPI with N8N to orchestrate automated workflows, voice inputs, and multi-step agent logic. In these sessions, I demonstrate real-time debugging, pipeline assembly, and tool-call orchestration.
Video Walkthroughs
Integrating VAPI with N8N – Session 1
Integrating VAPI with N8N – Session 2
Integrating VAPI with N8N – Session 3
What I Demonstrated
- •Setting up an N8N workflow that captures and processes data through VAPI.
- •Connecting HTTP nodes, function nodes, and external APIs.
- •Building dynamic logic chains for automated task execution.
- •Dealing with common configuration and authentication pitfalls.
- •Designing scalable and traceable automation pipelines.
"These videos show my practical ability to integrate LLM tooling with workflow automation engines and to debug, reason, and design real-world automation systems."
Intelligent Customer Service Agent (Tax & Order Support)
A multilingual AI support agent designed to augment our internal team.
Key Capabilities
- •Connects securely to the Odoo ERP (MongoDB backend) to answer order inquiries.
- •Interprets complex legal documents (NDPA regulations, New Tax Acts) for compliance advisory.
- •Learns from human feedback to improve accuracy over time.
- •Includes detailed performance tracking, safety monitoring, and robust error handling.
Technical Notes
- Resolved SRV connection issues caused by Pydantic’s MongoDsn auto-append behavior.
- Migrated embedding stack from OpenAI → Gemini → OpenAI again due to quota failures.
- Implemented retry logic, exponential backoff, isolated reproduction scripts.
- Final stable stack: langchain-chroma, FastAPI server, OpenAIEmbeddings.
Embeddings & Vector Database Architecture
Built a robust vector search pipeline for retrieving documents and enabling context-aware reasoning.
Highlights
- •Designed self-contained VectorDB class accepting injected embeddings.
- •Debugged silent Windows crashes by: Running with python -u for unbuffered output, Using FakeEmbeddings for isolated tests, Renaming classes to avoid cached bytecode, Moving telemetry flags to top of script
- •Solved environment-specific conflicts by switching from chromadb to langchain-chroma.
"This architecture now supports stable semantic search, offline testing, and reliable updates."
API Integration & Model Migration Framework
A flexible architecture allowing migration between LLM providers depending on cost, quotas, or speed.
Features
- •Plug-and-play embedding backends
- •Graceful fallback when models throttle, fail, or hit quotas
- •Detailed monitoring of error patterns (429, network errors, timeouts)
- •Safe rollback system to previous stable model configurations
"This system ensures uninterrupted inference even during large-scale workloads."
FastAPI Production Backend
A fully structured backend powering the AI agent.
Key Details
- •Pydantic validation for all environment vars
- •Modular routing and dependency injection
- •Development hot-reload and clean production deployment
- •Logging, error-handling middleware, request lifecycle instrumentation
"This backend architecture now acts as a reusable template for new projects."
What I'm Looking For
I want to work with a team building agents that autonomously plan, execute, and optimize real-world tasks. I am especially excited about marketing automation, self-improving agents, and systems that blend LLM reasoning with enterprise-grade reliability.
My Ideal Role:
- ▹Build multi-step agent workflows
- ▹Integrate them with real platforms (Meta, Google Ads, TikTok)
- ▹Design feedback loops and memory systems
- ▹Ship production-quality AI systems that deliver measurable results