You're seeing this page as if you were . The main menu is still yours, though. Exit from immersion
Chris AbbottCA

Chris Abbott

AI Infrastructure Architect | RAG & Classification

€1,158/day
London, GB
15+ years

Average response time: 1 hour

About Chris

I build things that sit at the intersection of advanced data infrastructure and AI systems. Not dashboards. Not wrappers around APIs. Infrastructure.

Most recently: TNT Research, where I've built a temporal computing kernel and Dewey — a hierarchical classification engine with 50+ taxonomies, 13,000+ cross-domain references, and a multi-timestream architecture that lets organisations customise canonical standards without forking them. Dewey outperforms standard Vector RAG/Graph RAG/BM25/LLM re-ranking pipelines in testing. I designed, architected, and built the entire system as a solo founder working with AI-assisted development.

Before that: American Express, helping maintain and build financial systems at scale.

Before that: 14 years running DetectRight, a device detection engine used by enterprises globally. Same DNA — taking a classification problem the industry solved with brute force (regexp matching, static databases) and replacing it with structural intelligence.

I also wrote five published books on video game sound technology and ran a Kickstarter campaign that raised over £500K. Eclectic is an understatement.

What I actually bring to client work: I execute projects that would normally require a multi-disciplinary team. Research, system design, prototyping, feasibility analysis, business model insight — in one person. As temporary CPO for a real-estate startup, I designed back-end infrastructure, built a cost simulator for competitive analysis, mocked up an AI-powered interface that surfaced new monetisation opportunities, and advised on trademark issues — in a month.

I'm at my best when the problem hasn't been solved before and requires someone who can think structurally across domains while still writing the code. If you need a contractor who colours inside the lines of an existing spec, I'm probably not your person. If you need someone who can see what you're actually building and make it significantly better than you planned, let's talk.
  • English

    Native or bilingual

Remote only
Primarily works remotely

Experience

  • TNT Research
    Founder and Architect
    January 2025 - Today (1 year and 5 months)
    London, United Kingdom
    TNT stands for "Temporal Native Technology." I build custom RAG infrastructure that solves problems most implementations quietly ignore.
    Standard RAG retrieves from whatever you fed it and knows nothing about the world outside that corpus — except by accident. When it works, it looks impressive. When it fails, it fails silently: wrong context, stale information, confident nonsense. Most teams discover this the hard way, usually after they've built their pipeline around an approach that can't be fixed without rearchitecting.

    I've spent 14 years in classification infrastructure (DetectRight — device detection for enterprises) and the last two years building TNT: a temporal computing kernel that treats time, context, and causality as structural primitives rather than afterthoughts. The first product from that work is Dewey, a hierarchical classification engine backed by 46+ curated taxonomies and 13,000+ cross-domain references. It gives your ingestion pipeline something no vector database or graph RAG setup provides: awareness of how your documents relate to the broader domain, not just to each other.

    What this means in practice: more accurate classification at ingestion, dramatically less hallucination downstream, and retrieval that understands temporal context — when something was true, not just that it appeared in a document.

    I work best with teams that have already built a RAG pipeline and found the ceiling — the point where adding more documents or tweaking prompts stops improving results. That's usually a structural problem, not a tuning problem, and it's what I solve.

    Python, TOML-based configuration, designed to integrate with existing stacks rather than replace them.
    Ingestion Pipelines LLM Python Contextual RAG Taxonomy Creation
  • High Technology Publishing
    Director
    PUBLISHING
    June 1997 - Today (29 years)
    Bromley, United Kingdom
    Although this company was created to hold music copyrights for old Commodore 64 game themes (as well as to release remix CDs based on them, and to hold live events), it is also a catch-all corporate vehicle for IT Contracting and other non-Device-detection-related activities.
  • 8-bit Symphony
    Founder
    July 2019 - Today (6 years and 11 months)
    Bromley, UK
    8-Bit Symphony is a venture with the goal of encouraging live performance of 8-bit music, with a special emphasis on the orchestral tracks featured in the first concert, and preserving 8-Bit music through protecting the publishing rights of 8-bit musicians for future generations.

Recommendations

Be the first to recommend Chris

Help this freelancer shine by sharing your experience working together.

These freelancer profiles also match your criteria

AgathaA

Agatha Frydrych

Backend Java Software Engineer

4.7

(3)

2

BaptisteB

Baptiste Duhen

Fullstack developer

4.6

(4)

5

AmedA

Amed Hamou

Senior Lead Developer

4

(2)

7

AudreyA

Audrey Champion

Web developer

4.3

(3)

4

Education

  • BSc Hons, Psychology and Statistics
    University of Surrey
    1992
    BSc Hons, Psychology and Statistics

Skill set (13)

Categories