Founder photo
Drop in at /public/images/rittick.jpg
About
Dr Rittick Barua, founder and principal of Nokshi Technology.
Nokshi Technology was founded in 2025, out of an AI delivery engagement with Rud Pedersen Group and a parallel research-engineering affiliation with a UCL surgical biotechnology group at the Royal Free Hospital. The founder is a chemical engineer by training: PhD from the University of Cambridge, MEng (First Class Honours) from University College London. Named co-author on a UK Government Department for Science, Innovation and Technology (DSIT) whitepaper on cyber-security risks to artificial intelligence (2024), and on the Jersey Finance guide to artificial intelligence in Jersey's finance industry (2024).
Why Nokshi exists
Engagements that conclude with a working system, not with a written deliverable about one.
Organisations approaching applied artificial intelligence tend to arrive at one of two terminal states: a written advisory engagement whose recommendations are never built, or a proof-of-concept whose architecture cannot be promoted to production without effectively rebuilding it. Nokshi Technology was founded to operate in the space between those two failure modes — engagements that produce a production-grade system, owned by the client, that the client's own team is able to operate after handover.
The studio is deliberately founder-led. Clients work directly with the principal, rather than through an associate pipeline. Where the scope of an engagement exceeds the capacity of a single engineer, we bring in collaborators with whom we have worked previously; the technical responsibility, however, remains with the founder.
The studio's typical engagements fall into three categories: European public-affairs and strategic-communications groups operating across multiple regulated jurisdictions; UCL-affiliated surgical biotechnology and clinical research groups requiring computer-vision or interpretable-modelling work as part of a published study; and UK financial-services firms, including those operating under Crown Dependency residency constraints. The common factor across the three is the presence of sensitive data, meaningful compliance constraints, and material consequences if the artificial-intelligence layer behaves incorrectly.
Background
Selected roles and engagements.
The entries below are selective rather than exhaustive — included because they inform how the studio works, not because they complete a chronology.
-
2025 — present
Founder and principal
Nokshi Technology Limited
Founded in 2025, following a year as AI delivery lead on the extended LibreChat deployment for Rud Pedersen Group and a parallel research-engineering affiliation with a UCL surgical biotechnology group at the Royal Free Hospital. The studio was established to take on further engagements of that character: production systems in regulated and research settings, where the consequences of the artificial-intelligence layer behaving incorrectly are material, under a consistent brand and with the technical responsibility resting with the founder.
-
2025 — present
Research engineering (affiliated)
University College London — Royal Free Hospital
High-throughput cell-spheroid segmentation in PyTorch. A U-Net architecture with classical baselines for comparison; validation intersection-over-union of 0.968 and Dice coefficient of 0.983. Applied computer vision in support of surgical biotechnology research.
-
2022 — 2024
Head of Applied Science
Grant Thornton
Led a generative-AI programme of approximately ten million pounds in committed value across five divisions, with twenty-five-plus production use cases, eight million pounds-plus in projected efficiency savings, two thousand-plus staff demonstrations, and the technical mentorship of seven junior data scientists.
-
prior
Earlier roles
Appian; research and consulting engagements
Low-code enterprise automation, including the Appian Bench Management product (approximately £1.1m of recognised value and a six-thousand-user deployment at the European Food Safety Authority); and applied research in scientific computing and physics-based modelling.
Published work
Named-author publications.
-
UK Government · DSIT · 2024
Cyber security risks to artificial intelligence
McCay, Al-Khalidi, Peng, Crossman-Smith, Barua
-
Jersey Finance · 2024
Guide to Artificial Intelligence in Jersey's Finance Industry
McCay & Barua
Method
Three phases, applied consistently across engagements.
The intention behind keeping the engagement shape consistent is twofold: to make the scope, deliverables, and handover criteria legible to the client from the outset, and to produce a system the client's own team can continue to operate once the engagement concludes.
A paid discovery phase of approximately two to four weeks, at the end of which the client receives a written architecture and delivery plan. The deliverable is the client's to retain whether or not the engagement proceeds further.
A fixed-scope build with defined milestones. Infrastructure, continuous integration, monitoring, and security review are treated as first-class scope rather than as activities appended after a working demonstration has been produced.
Handover is treated as a design property of the engagement. Documentation, environments, and architectural reasoning are written down in a form the client's engineers can read and challenge. An iteration retainer is available where useful, but a permanent dependency on the studio is not the objective.
Scope boundaries
Work the studio is unlikely to be the right home for.
- Long strategy engagements that conclude with a written deck rather than a working system.
- Proof-of-concept artificial intelligence that is decommissioned the moment the engagement ends.
- Per-seat licensing models or platform rent-seeking arrangements.
- Multi-supplier pitch competitions assessed primarily on slideware. The work that follows tends to be poorly served by them.
- Deep individual-contributor cloud-platform engineering. Where this is the principal need, we are happy to recommend specialists with whom we have worked.