RadTech
Book a call →

RadTech · AI & data · deep tech · products

Radical technology
that moves
drug discovery
forward.

RadTech turns frontier AI and data into systems that ship, for teams doing hard technical work. I also design and ship my own AI products end to end. Strategy through delivery, from someone who has run the programs and built the apps, not just advised on them.

Book a discovery call → See the work →
Programs run with
ENPICOMNovartis Data42GenialisThe Hyve

Selected work

Fifteen years across antibody discovery, FAIR data, and AI strategy. These are the engagements that moved real programs forward.

All work & code →
ENPICOM
Project Lead · 2025 → present
Antibody discovery platform. Leading cross-functional programs turning immune-repertoire NGS into therapeutic antibody candidates, FAIR-by-design, ISO 27001.
>€1M
programs led
Novartis Data42
FAIR consultant · core team · 2019 → 2021
Knowledge-graph governance. Embedded from inception to delivery. The governance design decides whether a billion-dollar data platform is actually queryable.
30+
stakeholders
Genialis
Program Lead · 2023 → 2024
AI biomarker discovery. Cancer-biomarker programs combining AI, ML, and NGS across multi-site clinical and computational delivery. Award-winning PM.
Award
winning delivery

How I help

Three engagement shapes, honest scoping, no retainer dance.

Services & process →
01

AI strategy & business cases

From hypothesis to funded program: opportunity sizing, build-versus-buy, and an AI roadmap your team and your board both believe. This is the work that decides where AI earns its keep and where it is theatre.

Scoped per engagement
DeliverableAI roadmap + board-ready business case.
02

FAIR data platforms

Linked-data and knowledge-graph architecture for regulated, data-intensive environments. Schema, vocabulary, ingestion, governance: FAIR by design rather than bolted on, so a model can actually query your data.

Fixed fee, quoted after scoping
DeliverablePlatform blueprint + architecture your team can execute.
03

Discovery & advisory

Embedded technical advisory for discovery, data, and product teams. Think of it as the translator between scientific intent and shipped systems, the senior technical voice your program has been missing.

Monthly retainer, ask for the rate
EngagementRetained, remote-first, monthly.

Talks & speaking

FAIR, AI in drug discovery, and knowledge graphs, for pharma and biotech audiences.

Full bio & talks →
2026-05-15
Booth + panel · PEGS Boston 2026
PEGS Summit, Boston · immune-repertoire NGS conversations.
Booth lead
2026-04-23
Data Governance in Pharma: Navigating the Digital & AI Revolution
Pistoia Alliance · European Life Science Informatics Forum Benelux (5th) · GSK Wavre.
Speaker + host
2024 → 2025
Invited talks: FAIR, AI in drug discovery, knowledge graphs
Pharma + biotech partner sessions across Europe. Full list on request.
Speaker
SS

Sukhi Singh, founder of RadTech. I work on the data platforms behind antibody discovery, the layer between a sequencer and a decision, and I build my own AI products on the side. A scientist's rigour with a builder's restlessness, which is roughly how an engagement tends to feel.

PhD Bioinformatics · Max Planck Wharton Entrepreneurship Pistoia Alliance 80+ countries
More about Sukhi →

Field notes

Writing · proof of work
2026-05-20 PEGS Boston 2026: Design Is the Easy Part Now Antibody Eng.
2026-04-24 VHH Humanization Is Mostly a Search Problem, Not a Grafting Problem Antibody Eng.
2026-04-21 Data Governance: Notes from the Pistoia Alliance Benelux Forum #5 Pharma + AI

Let's see if RadTech is the right call.

A 30-minute discovery call, free, no pitch. If we are a fit, you get a fixed-scope proposal within 48 hours.

Book a discovery call → Other ways to reach me →
contact@radtech.nl · /in/vanbug · Amsterdam, NL

// Portfolio · engagements & case studies

Selected portfolio

Fifteen years across antibody discovery, FAIR data, genome assembly, and AI strategy. A short list of the engagements that taught me the most. For the products and tools I build myself, see Projects.

Project Lead · 2025 → present
ENPICOM
Antibody NGSFAIRISO 27001

Antibody discovery platform

Leading >€1M cross-functional programs turning immune-repertoire NGS into therapeutic antibody candidate shortlists. Drive the product roadmap with scientists, engineers, and external pharma partners. Multi-million-euro budgets, FAIR-by-design, ISO 27001 environment.

Principal Consultant · 2024 → present
RadTech
AI strategyFAIRBiopharma

AI & data strategy for biopharma

Advise pharma and biotech organisations on AI, data governance, and FAIR platforms, from business case through delivery. Engagements span early opportunity sizing to multi-million-euro programs, bridging scientists, engineers, and leadership.

Program Lead · 2023 → 2024
Genialis
AI biomarkersCancerGDPR + HIPAA

AI biomarker discovery program

Led cancer-biomarker programs combining AI, ML, and NGS. Coordinated KOLs, CROs, and senior leadership. Award-winning project management across multi-site clinical and computational delivery.

FAIR consultant · core team · 2019 → 2021
Novartis Data42
Knowledge graphsFAIRTop 5 pharma

Data42 knowledge graph

Embedded on the Data42 program from inception through delivery as a FAIR consultant and core team member, working with 30+ stakeholders across business, science, and engineering. This is the governance design that decides whether a billion-dollar data platform is actually queryable.

Bioinformatician · 2018
HZPC Research
Genome assemblyPolyploid

First tetraploid potato genome assembly

Contributed to the first published tetraploid potato genome assembly, a hard polyploid problem with direct breeding-program implications. Published, cited, productionised.

Have a program that needs to ship?

Roadmap, platform blueprint, or embedded advisory: a 30-minute call decides which fits.

Get in touch →

// Projects · things I build

Built, not just advised.

Outside client work I design and ship my own products and tools, usually solo and end to end, from the research spike to running code. A selection of what I have been building.

Pourtrait · coffee
Product · solo founder · 2026 · in development

Pourtrait: rate the barista, not the cafe

A discovery app for specialty coffee that scores the person who made your drink, not just the venue it came from, with live shift-aware search for who is actually working right now.

  • Location-aware discovery of nearby baristas and venues, built on a live NL specialty-coffee dataset.
  • Claim-your-profile identity and reputation: consent-gated scoring so a barista's rating only shows once they've claimed it.
  • Trust and safety built in from the start: blocking, reporting, moderation, and duplicate-profile detection.
React NativeSupabasePostGISTrust & safety
In development · Android beta
Kawa · routing
Product · solo founder · 2026 · in development

Kawa: a community road-intelligence layer for motorcyclists

An AI route planner that finds the good roads, not just the fast ones, and quietly learns which roads are actually worth riding from the community that rides them.

  • Plans curvy, scenic, toll- and highway-aware routes on open map data, then exports turn-by-turn GPX for offline navigation. No API key, no build step.
  • Builds a living road-quality map from anonymous, blurred road geometry: privacy-first by design, with no tracking of individual trips.
  • Private and group ride sharing, a lightweight rider identity, and a merit system that rewards mapping and rating real roads.
AI planningOpenStreetMapLeafletGPX / PWAPrivacy-first
Working v1 · private beta
Kawa · ride atlas
Interactive studies · 2026 · vanilla JS, zero dependencies

Kawa Ride Atlas: learn the machine, then the physics

Two interactive plates that teach a motorcycle from the inside out, built as design studies for the Kawa brand.

  • Exploded parts atlas. Pull the bike apart component by component; click any part to read what it does and why it is shaped that way. 23 parts across 6 systems.
  • Lean & cornering simulator. Cornering as a live physics problem: pick a body position, change speed, radius and grip, and watch the lean angle and force vectors fall out of the maths in real time.
Interactive vizReal-time physicsInline SVGZero dependencies
Prototype
agent-ken · orchestration
AI tooling · personal · 2026 · Python

agent-ken: a parallel, self-reviewing agent orchestrator

A dependency-aware kanban that takes a single brief, turns it into a plan, and ships it with a fleet of AI agents, without me babysitting each step.

  • Decomposes a task into subtasks with explicit dependencies, then runs the independent ones in parallel waves.
  • Gates every subtask through an automated reviewer and only marks work done once it passes: quality built into the loop.
  • Generates live kanban, mind-map and flow visualisations of the whole run as self-contained HTML dashboards.
AI agentsOrchestrationPythonLLM review loop
Personal tool

Code & tools

Smaller things I have built and use weekly.

new-post.py

Python

Live news + literature digest piped through an LLM into a finished blog post, four LinkedIn variants, a Bluesky cross-post and a newsletter draft. A nine-step pipeline with stage banners.

business-week.py

Python · yfinance

Weekly antibody-business analyst pipeline: market prices, scoped news and RSS, LLM batch categorisation, and a generated 600 to 800 word analysis.

fair-cookbook

FAIR · standards

Contributions to the FAIR Cookbook on biopharma data practices: pseudonymisation, linked-data modelling, governance patterns.

faircookbook.elixir-europe.org →

projectAnalytics

Python · uv

Self-contained HTML dashboards generated from project, time-tracking and HR data sources: executive view, analytics, timeline, budget and resourcing for reporting and capacity planning.

Want something like this built?

Most of these started as a weekend research spike. If you need one shipped properly, that is the day job.

Get in touch →

// Services

I turn expensive data bottlenecks into shipped systems.

AI and data strategy for teams doing hard, technical work. I have run cross-functional programs at ENPICOM, Novartis Data42, Genialis, and The Hyve. Three engagement shapes, honest scoping, and a straight answer on fit before we talk fees.

01

AI strategy & business cases

From hypothesis to funded program: opportunity sizing, build-versus-buy, and an AI roadmap your team and your board both believe. The work that decides where AI earns its keep in your roadmap, and where it is theatre.

Scoped per engagement
DeliverableAn AI roadmap and a board-ready business case.
02

FAIR data platforms

Linked-data and knowledge-graph architecture for regulated, data-intensive environments. Schema, vocabulary, ingestion, and governance, designed FAIR by design rather than bolted on, so a model can actually query your data.

Fixed fee, quoted after scoping
DeliverableA platform blueprint and architecture diagram your team can execute.
03

Discovery & advisory

Embedded technical advisory for discovery, data, and product teams. Think of it as the translator between scientific intent and shipped systems, the senior technical voice your program has been missing.

Monthly retainer, ask for the rate
EngagementRetained, remote-first, monthly.

How an engagement runs

  1. 01Scoping callFree, thirty minutes. I ask the questions that decide whether you actually need an outside specialist.
  2. 02Fixed proposalWithin 48 hours: scope, deliverable, fee, and what I need from you. No retainer dance.
  3. 03ExecutionI work directly with your lead. Weekly written updates. Final deliverable handed over, no consultancy-speak.
  4. 04HandoverYour team owns it. I'm available for thirty days of follow-up questions at no charge.

Roadmap, platform, or advisory?

A 30-minute discovery call decides which fits, and whether you need an outside specialist at all.

Book a discovery call →
Sukhi Singh, founder, RadTech

// About the founder

Sukhi Singh

Founder of RadTech. I work on the data platforms behind antibody discovery, the layer between a sequencer and a decision, and I design and ship my own AI products on the side. Currently also Project Lead at ENPICOM in Den Bosch.

SS

I write and build the things that have to be said but rarely get said inside pharma BD decks. Call it the unglamorous middle layer: how data is shaped before a model sees it, which platform decisions kill a discovery program quietly, and where AI actually buys you something versus where it's theatre.

Outside the work: I've travelled to more than 80 countries, ride motorcycles, DJ, play ultimate frisbee, and hike whenever the calendar allows.

ENPICOM · Project Lead Amsterdam / Den Bosch Pistoia Alliance PhD · Max Planck
Get in touch →

// Experience

Sep 2025 → present
Project Lead
ENPICOM
Lead >€1M cross-functional antibody discovery + analytics programs. R&D, AI, product roadmap. FAIR by design. ISO 27001.
Mar 2024 → present
Principal Consultant
RadTech
AI & data strategy, FAIR, linked data. Global partnerships across Asia and LATAM.
Mar 2024 → Feb 2025
AI Program Manager
Assal Energy
Built and deployed AI & GenAI business cases for renewable energy with ministries, PE, and ops teams.
Apr 2023 → Feb 2024
Program Lead, AI Biomarker Discovery
Genialis
Cancer biomarker programs combining AI, ML, NGS. FAIR, ISO 27001, GDPR, HIPAA compliant.
Mar 2021 → Apr 2023
Senior Applied Scientist & Alliance Manager
Genialis (Boston)
Biopharma consulting on data science, bioinformatics, AI/ML, multi-modal analysis for precision health.
Nov 2019 → Mar 2021
FAIR Data & AI Consultant
The Hyve
Advised Top 15 pharma on knowledge engineering, OMOP/CDISC mapping, RWE. Embedded on the Novartis Data42 knowledge graph as core team member.
2010 → 2019
Bioinformatician
HZPC · Max Planck · TU Dresden
First tetraploid potato genome assembly. Epigenetics. Editor at biostars.org.

// Education

2024 → 2025
Entrepreneurship Specialization
The Wharton School
Grade 90%. Pitch development, venture financing, lean scaling, team building.
2011 → 2016
PhD, Bioinformatics
Max Planck Society
Magna cum Laude (1.3). Six papers on H3K4 methyltransferases in mESCs.

// Selected publications

FAIR Cookbook: community contribution on biopharma data practices.
BOD1L is required for DNA replication fidelity (Cell Reports, co-author).
Set1 complex stabilizes H3K4 dimethylation at pluripotency genes (Mol Cell).
Kmt2b regulates transcriptional bursting during ES differentiation (Genome Biology).
Oct4 + Mll2: maintenance of bivalent chromatin in mESCs (Nature Comms).
Full list on Google Scholar (y62wfS8AAAAJ)

Speaking & media

Selected, chronological.

2026-05-15
Booth + panel · PEGS Boston 2026
PEGS Summit, Boston · ENPICOM booth lead, immune-repertoire NGS conversations.
Booth lead
2026-04-23
Data Governance in Pharma: Navigating the Digital & AI Revolution
Pistoia Alliance · European Life Science Informatics Forum Benelux (5th) · GSK Wavre.
Speaker + host
2024 → 2025
Invited talks: FAIR, AI in drug discovery, knowledge graphs
Pharma + biotech partner sessions across Europe. Full list on request.
Speaker

// Contact

Let's see if RadTech is the right call.

A 30-minute discovery call, free, no pitch. I ask the questions that decide whether you actually need an outside specialist. If we are a fit, you get a fixed-scope proposal within 48 hours.

Book a discovery call →

// Direct

Email · contact@radtech.nl
LinkedIn · /in/vanbug
Based · Amsterdam, NL

RadTech works with teams doing serious technical work across life sciences, deep tech, and data-intensive industries, plus my own AI products on the side. If we are not the right fit, I'll say so on the call.