AI Lead

About Disseqt AI

Disseqt is the Assurance Layer for Enterprise AI Operations. One platform covers the full deployment lifecycle: pre-production testing, runtime policy enforcement, continuous post-deployment monitoring and compliance reporting.

AI agents are non-deterministic. They hallucinate. They produce unexpected outputs. They behave differently in production than in any test environment. They drift. They degrade. Enterprises deploying AI agents at scale have limited visibility into what those systems are doing after go-live. Legacy GRC tools were built for predictable software and cannot monitor what changes continuously.

Disseqt closes that gap. With regulations like the EU AI Act and ISO 42001 now mandating full lifecycle AI governance, enterprises using Disseqt are winning regulated deals, faster and with less risk.

About the role

AI models already power many parts of our product. You will build Disseqt's AI and research team from the ground up and create the system that turns data into trained, evaluated and served models, for text first and then for multilingual, image, audio and video.

You will be the first senior hire in this team. You will set the model strategy, hire the team, decide which models we build on, and make sure every model we ship is measured, reproducible and legally clean. Three engineers already on the team will move into your organisation.

What you will do

First 3 months

• Hire your first leads: a Data lead, a Research scientist for evaluation, and an MLOps / training infrastructure engineer.
• Start building base models and productized processes  for training and fine-tuning.
• Write the 12-month model roadmap and get it signed off by leadership.
• Guide the three existing engineers through their first tasks: a data inventory, a model and cost inventory, and draft test sets.

Months 4 to 6

• Complete Phase 1 hiring (about 10 people).
• Take  models through the full loop: versioned data, training, evaluation, serving behind an API, and production results logged back into datasets.
• Set up the release process. No model ships without passing evaluation, and you sign off releases together with the evaluation lead.

Months 7 to 12

• Replace third-party model calls in the product with models we own, where they meet quality, latency and cost targets.
• Extend the same pipeline to priority languages and to image, then audio.
• Move the team from a flat structure to three leads (Data, Modeling, Platform) and grow it to about 22 people.
• Launch the first external model API and the first public benchmark, with the AI product manager and evaluation team.

Ongoing

• Own decisions on base models, training data rights and model licensing, working with legal.
• Run a monthly model review covering quality, latency and cost for every model in production.
• Run a monthly IP review of invention disclosures and dataset provenance.
• Represent the team's work externally through benchmarks, technical writing and customer conversations.

The team you will build

• Data: Data lead, annotation and human review manager, associate data engineer.
• Modeling: senior ML engineer (fine-tuning lead), ML engineer for classifiers and small models, ML engineer for multilingual.
• Evaluation and benchmarks: research scientist for evaluation, associate evaluation engineer.
• Platform: MLOps / training infrastructure engineer, inference engineer, associate MLOps engineer.
• Product: AI product manager.

Later hires include vision and speech engineers, a decision scientist, an API engineer, a benchmarks research engineer and developer relations.

What you bring

Must have

• 8+ years in machine learning, with at least 3 years leading ML teams.
• A track record of taking models from data to production and owning their quality, latency and cost after launch.
• Hands-on experience fine-tuning open-weight models (for example SFT or LoRA), distilling large models into smaller ones, and training classifiers.
• Experience building evaluation practice: held-out test sets, human agreement, regression testing and release criteria.
• Working knowledge of ML infrastructure: training pipelines, experiment tracking, model registries and model serving.
• Experience hiring and developing engineers, including mentoring junior engineers.
• Sound judgment on data provenance and model licensing.

Nice to have

• Experience in AI safety, content moderation, trust and safety, or security ML.
• Experience with multilingual models, especially for low-resource languages.
• Experience with vision or speech models.
• Experience shipping models as external APIs.
• Publications, open-source work or public benchmarks.
• Startup experience.

What success looks like

• By month 3: first three leads hired, base models chosen, 12-month roadmap approved.
• By month 6: Phase 1 team in place; one model working end to end through data, training, evaluation, serving and logging.
• By month 12: owned models in production replacing third-party calls; image models live and audio in progress; external API in beta; first public benchmark published.