Engineering Frontier
AI Systems from Concept to Production.
Senior AI Developer and former Enterprise Product Owner. I build end-to-end autonomous multi-agent pipelines, Vision-Language processing systems, local GPU diffusion engines, and low-latency edge AI with deep production rigor.
SPECIALIZATION Multi-Agent, VLMs, Local Diffusion, Edge AI
TECH Python, TypeScript, Rust, C++, FastAPI, Docker
The Rare Hybrid: Deep AI Builder + Enterprise Product Owner
Most AI projects fail at the seam between engineering and business delivery. Having managed \$4M+ ARR enterprise client backlogs at Hippo Video and built interfaces at Zoho, I build AI systems designed from day one to deliver quantifiable business value.
Agentic AI & LLMs
Going beyond single prompt wrappers. Designing autonomous orchestrators, multi-agent critique loops, self-correction nodes, and privacy-preserving PII redaction pipelines.
- Multi-agent copy & art generation
- Autonomous critique & human-in-the-loop
- RAG & hybrid OCR + LLM parsing
Computer Vision & VLMs
High-throughput multimodal perception. Local GPU inference pipelines with Stable Diffusion Turbo, Qwen2.5-VL batch culling, and facial recognition for real-world automation.
- Local GPU SDXL Base & Turbo studio
- Qwen2.5-VL photo & video culling
- Instant facial cluster & selfie matching
Production & Edge Engineering
Building systems that scale cleanly. Microservices with Python FastAPI, enterprise RBAC security, Google OAuth 2.0, low-latency Rust recording, and C++ offline voice edge assistants.
- Role-Based Access Control (RBAC) & OAuth
- High-performance Rust screen recording
- C++ offline voice & MQTT Kuksa VSS
End-to-End Technology Stack
Production-tested frameworks, local inference runtimes, edge protocols, and full-stack architecture.
Generative AI & LLM Architectures
Computer Vision & Multimodal Media
Edge AI & High-Performance Systems
Backend, Security & Cloud Platforms
Flagship Public AI Projects
Explore open-source implementations showcasing computer vision, LLM inference, edge vehicle simulators, and document intelligence.
VideoPrompt Studio
Deployed live at yourstoryas.video. End-to-end AI cinematography & video prompt engineering studio. Features a multi-stage generation pipeline (Story → Scenes → Shots → Frame enhancement), 8-axis camera/lighting gear controls, Cloudflare R2 presigned object storage, and DeepInfra model routing.
PromptBuilder
Deployed live at octoapt.com. Production prompt engineering and creative synthesis engine built for structured prompt compilation, reference styling, model parameter calibration, and automated model-tuned outputs.
ResumeMatchAI
Solves candidate screening fatigue for recruiters. Features candidate Magic Link workflows, hybrid OCR + LLM matching pipeline, automated PII redaction, and recruiter ranking dashboards.
EdgeDrive Simulator
Physical vehicle telemetry simulator integrating an offline C++ AI Voice Assistant, camera-based ADAS alarms, and Eclipse Kuksa.val Databroker using COVESA VSS over an MQTT messaging pipeline.
Alt-Tag Generator
Image accessibility enhancer generating rich, descriptive metadata and WCAG-compliant alt text for local and web images using vision-capable multimodal LLMs.
Photo Culling Pro
Web-based photo culling solution with AI-powered image quality scoring, face detection, expression analysis, duplicate clustering, and batch photographer review workflows.
PDF Vision Image Extractor
Document processing engine that parses PDFs, detects embedded graphical figures, computes precise bounding boxes, requests operator confirmation, and indexes assets into database storage.
Meta-AI Prompt Analyser
Specialized prompt evaluation and syntax analysis toolkit designed to benchmark LLM instruction adherence, token efficiency, and prompt structuring.
Billing & Timesheet Suite
Transparent client-versus-company invoicing software mapping active engineer timesheets directly to audited invoices for seamless client billing trust.
Production Systems & Custom AI Engines
A look into proprietary production systems, custom agent architectures, and local GPU pipelines engineered for high-performance enterprise workloads.
Autonomous Multi-Agent Ad & Creative Generator
Designed an autonomous multi-agent pipeline where users describe a product, and specialized agent nodes collaborate to generate production-ready advertising collateral across distinct print and digital formats (A3 newsprint, magazine spreads, social hero banners).
Local GPU Studio for Zero-Cloud-Cost Generation
Engineered an offline-first, high-throughput GPU studio running quantized diffusion models on consumer hardware. Dramatically slashed API billing by executing latent synthesis locally with zero external API calls.
VLM Photo Culling & Instant Selfie Face Matcher
Revolutionized event and wedding photography workflows. Photographers upload thousands of raw exposures; an automated Vision-Language Model culls out blinks, blur, and poor framing, while an embedded facial vector index matches guests via a single selfie.
Multilingual Audio-Video Synchronization & Podcast Pipeline
Automated audio extraction, transcription, multilingual dubbing, and video subtitle burning. Synchronizes phonemes and word-level timestamps using modern neural speech engines and programmatic FFmpeg rendering.
Professional Milestones
A solid track record of leading product initiatives, managing enterprise ARR, and delivering engineering excellence.
AI Product & Workflow Consultant
Independent Practice Oct 2023 – PresentArchitecting and shipping end-to-end AI applications, multimodal vision agents, and low-latency automated workflows for clients.
- Shipped & Deployed VideoPrompt Studio (yourstoryas.video) & PromptBuilder (octoapt.com) for automated video prompt compilation and creative generation.
- Shipped ResumeMatch AI: Privacy-first resume screening platform with OCR + DeepSeek-V3 and automated PII redaction.
- Engineered Agentic Creative Platform: Multi-agent pipeline with auto-critic loops and human-in-the-loop controls.
- Built Make Way to Save a Life: Google Maps-integrated emergency operations platform with real-time ETA tracking across ambulance, police, and hospitals.
Product Owner & Customer Operations Lead
Hippo Video Apr 2021 – Sep 2023Led customer-facing backlog, ran daily scrums, and bridged enterprise customers with core engineering.
- Personally owned 30+ active high-value enterprise accounts contributing to $4M+ ARR.
- Led a 4-member customer operations team managing 20,000+ accounts (192 Enterprise / Mid-Market clients).
- Analyzed user telemetry (Pendo, Hotjar) to eliminate feature friction and fuel Product-Led Growth (PLG).
- Optimized outsource vendor and third-party SaaS contracts, driving significant cost reductions.
Senior UI Developer → Product-Engineering Bridge
ZOHO Corp (Adventnet) Oct 2005 – Aug 2012Led UI engineering for OpManager (flagship network monitoring product), bridging backend engineers with frontend designers.
- Architected AJAX/JSON live-data streaming UI components for high-throughput network monitoring dashboards.
- Eliminated cross-browser UI breakages and established company-wide modular frontend standards.
Let's Build Something Exceptional
Interested in collaborating on autonomous multi-agent pipelines, local GPU model deployment, computer vision systems, or engineering leadership? Reach out through the channels below.