×
Jetty
Managed infrastructure for agentic AI: sandboxed execution, real-time tracing, and self-evaluation loops
Last verified August 17, 2026 · Updated daily
What Jetty is building
Jetty is building infrastructure for agentic AI workflows across three primitives that solve for the same underlying problem: agents that ship reliably, not just run. The core concept is the runbook: a plain markdown file that gives an AI agent its job, its definition of done, and a method for checking its own work before calling it finished. Skills (the instructions) plus standards (the definition of done) equals a runbook. That equation is the product thesis in five words. An agent running against a runbook does not just execute a prompt. It generates output, verifies that output against defined criteria, retries if anything fails, and escalates to a human when it cannot resolve a failure after a fixed number of attempts. The agent definition lives as a markdown file in your repo, not as a graph in a framework or a config inside a vendor's console. You own it. Any model runs it. On top of runbooks, Jetty provides sandboxed execution environments (isolated workspaces where agents run without touching production systems), real-time tracing from instruction to output (every decision in the chain is logged), and evaluation loops that ingest traces, run assessments, and produce pull requests with verified improvements. Jonathan's Substack post "Generation Got Cheap. Verification Didn't." is the longest and most rigorous public articulation of the problem: as token costs collapse 30-80%, teams automate more tasks without expanding their verification capacity. The fraction of output that is actually verified shrinks with every new task the model takes on. Jetty is the verification infrastructure that closes that gap.
Why this matters
Jonathan's framing for the market timing is precise and worth understanding. He cites a 2026 MIT paper by Catalini, Hui, and Wu that formalizes what practitioners already feel: the cost to generate AI output is collapsing, the cost to verify it is not. The authors call the expanding zone between those two curves the Measurability Gap: the growing share of tasks where machines can cheaply generate output that humans cannot affordably verify. Jetty sits inside that gap as the tool that makes verification scalable. This is not a niche problem. Gartner projects 40% of enterprise applications integrating task-specific agents by the end of 2026. IDC projects AI copilots embedded in 80% of enterprise workplace applications in the same timeframe. Every one of those deployments faces the same question: how do you know the agent is working? How do you catch a regression when you swap models? How do you prove compliance to a regulator when the agent made a decision that affected a patient or a financial account? Current answers are: manual spot-checking, vibe-based dashboards, and crossing fingers. Jetty's runbook-plus-evaluation infrastructure is the systematic alternative. The Mila affiliation is also meaningful for the market. Mila is one of the two or three most productive AI research institutes in the world: Yoshua Bengio, one of the three people who won the 2018 Turing Award for deep learning, is its scientific director. Being embedded in the Mila Ventures space puts Jetty at the centre of a research-to-deployment pipeline that no other city in Canada has.
Open roles at Jetty
4 positions we're tracking. Roles are re-checked daily and removed when filled.
Founding Engineer (Agent Infrastructure / Evaluation Systems)
First seen 3 months ago
GTM Lead / Head of Partnerships (Enterprise AI)
First seen 3 months ago
Product Designer / Design Lead
First seen 3 months ago
Operations Lead
First seen 3 months ago
Know when Jetty is hiring before anyone else
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching Jetty
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+Hiring outlook
Very High. Press release states the round will be used to accelerate product development, expand the engineering team, and support enterprise traction. The company page explicitly states they are actively hiring across product, GTM, engineering, and operations. No ATS, no aggregator listings. Cold email only.
Working at Jetty
Since , Jetty has built Managed infrastructure for agentic AI: sandboxed execution, real-time tracing, and self-evaluation loops. The team is now <50 people. For a AI company this size, the reality is broad remit, direct access to founders, and equity that still means something if the company works out.
The majority of roles are in Montreal, Canada.
How to actually get hired at Jetty
Why applying the normal way doesn't work
At <50 people, Jetty has no recruiting team. Your application lands with a founder who is also running sales, product and payroll. The obstacle isn't a queue or an ATS, it's being seen at all. Cold outreach outperforms the form here, consistently.
Who to contact at Jetty
Founding team
Jonathan Lebensold's career moves in a specific pattern: ship something, then go understand why it breaks. He co-founded Paradem, a Montreal software consultancy that built line-of-business applications for startups and enterprises, and during that period wrote the React Native Cookbook for O'Reilly, one of the earliest comprehensive guides to cross-platform mobile development. Technical book writing at that level requires explaining a system end-to-end to someone who has never seen it with enough clarity to survive scale. He then pivoted hard into research: completing a PhD at McGill University and Mila under Borja Balle (now at Google DeepMind, one of the leading differential privacy researchers in the world) and Doina Precup (one of the most cited reinforcement learning researchers globally, co-director of Mila's Montreal office). His doctoral work focused on differential privacy, privacy-preserving machine learning, and generative model accountability: the formal discipline of proving that an ML system did what you permitted it to do, not just what you wanted it to do. He then worked as a visiting researcher at Meta AI on privacy-preserving ML and at Reliant AI on AI reasoning before founding Jetty. His Google Scholar profile has 1,331 citations. He is also a contributor to OpenMined's PySyft, implementing websocket infrastructure for distributed privacy-preserving ML, which is how he entered the research community before the formal PhD. His Substack, Ground Truth, is the most substantive window into his thinking: the post "Generation Got Cheap. Verification Didn't." is a 2,500-word rigorous case, citing an MIT paper on the Measurability Gap, that constructs the intellectual argument for why Jetty exists. He appeared on Nick Taylor's Streams to discuss agent hill-climbing and evaluation. His X bio reads: "AI has an evaluation problem and I'm trying to fix it." That is the product thesis in nine words. He books his own demos via Calendly. His family maintains a shared website at lebensold.ca where his personal page sits alongside those of Esther, Julian, and Suzanne. An early consultancy bio notes he enjoys baking apple pie. His academic supervisors are two of the most rigorous researchers Mila has produced. Both details are real and neither is a contradiction.
Ezra Hopkins is the UX and design co-founder at Jetty, and his path to that role runs through an unusually broad set of contexts. Earlier in his career he was a project lead at the Baha'i organization, contributing to initiatives around global unity, the harmony of science and religion, and international cooperation. That is not a common background for a software product designer, and it is not irrelevant: working in a governance-oriented, values-driven international organization builds intuition about how people make decisions under uncertainty and how information needs to be communicated to people who are not technical specialists. He then moved into design and web development professionally, working at TimeZoneOne, an international creative communications agency, designing web interfaces and applications using a ColdFusion-based CMS, and at Terabyte Interactive, building web pages with a .NET-based CMS while managing client relationships and subcontractor coordination. His design work is grounded in enterprise software interfaces: complex, information-dense, built for operators who need clarity under pressure, not for consumers who need delight. He joined Jonathan at Paradem as UX Lead, where the same brief applied: line-of-business applications that had to work reliably for people who could not afford to misread an interface. At Jetty, that problem is sharpened. The trace viewer, the escalation notice, the runbook editor, these are products that non-technical hospital administrators or financial operations leads will use to determine whether to trust an agent's output. If those surfaces are confusing, the verification problem does not get solved. Ezra is the co-founder responsible for making sure they are not confusing. His public profile is concentrated on LinkedIn and design portfolio work rather than social media or writing.
What to show them
Jetty's core technical product is the evaluation loop: an agent that runs, checks its own work against defined criteria, and retries or escalates when it fails. Building that reliably at scale, across multiple models and agent frameworks, while maintaining the tracing layer that makes every decision auditable, is the central engineering problem. Jonathan has a research background in privacy-preserving ML and AI safety; the engineer he hires for this role will be someone who thinks about agent behaviour with the same rigour. Core skills: Python, LLM agent evaluation frameworks, sandboxed execution environments (Docker, containers, isolated runtimes), distributed tracing systems (OpenTelemetry or custom), agent framework internals (LangGraph, AutoGen, or custom), CI/CD integration, GitHub Actions. Proof of work: Build a minimal evaluation harness for a two-step agent workflow: the agent executes a task, runs a self-check against explicitly defined success criteria, and either passes or retries with an explanation. Publish the code. Write a README explaining your choices for the success criteria schema and the retry logic design. The design of the criteria schema is the proof of engineering thinking here.
A cold email that works at Jetty
What Jetty screens for
AQC Capital and Hidden Layers Capital co-leading a pre-seed for a founder with a McGill/Mila PhD, Meta AI credits, and a published thesis on privacy-preserving ML is a research-credibility bet. They are backing the specific intellectual background that produced the runbook thesis, and they expect Jonathan to hire people with equivalent rigour. Hidden Layers Capital's name is itself a signal: neural network hidden layers are where the non-interpretable computation happens. A fund named after that concept is specifically betting on the infrastructure that makes AI systems more interpretable and accountable. The Akinox strategic investment is the most interesting signal in the cap table. Akinox builds digital health coordination platforms for hospital networks across Quebec and Canada: patient flow, care coordination, and inter-hospital communication infrastructure for the public health system. Their check is not a financial bet. It is a customer signal. Akinox is deploying AI in environments where error has regulatory and patient safety consequences. Jetty's runbook and evaluation infrastructure is exactly what a health IT platform needs before it can put AI agents into hospital workflows. Healthcare and government are the first and most urgent enterprise verticals. MLCommons is the second strategic signal. MLCommons is the industry consortium that creates standardized benchmarks for AI systems: its members include Google, Meta, Intel, Nvidia, Microsoft, and most major AI research organizations. Its participation in Jetty's round means Jetty is being positioned not just as a product but as governance infrastructure. MLCommons cares about standards. Jetty's open-source tools for model provenance and AI governance are designed to connect to those standards. The stated use of capital is explicit: accelerate product development, expand engineering, support enterprise traction. Three parallel mandates. Apply: apply@jetty.io
Customize your CV for the Jetty role. Matching the job description language helps clear ATS filters.
Don't make these mistakes
Enthusiasm about AI agents generally. Jonathan has been building and researching AI accountability for years. He will hear "I'm excited about the agentic AI space" as noise. Lead with the specific verification failure you have witnessed or the specific technical work you have done. The product thesis is precise. Your outreach should be too.
Mistakes that kill Jetty applications
Generic CVs stand out at a <50-person company — and not in a good way. Fastest path to rejection.
Skip 'I'm looking for...' — start with Jonathan's framing for the market timing is precise and worth understanding and your specific angle on solving it.
Applying and waiting = silence. Follow up at day five — it roughly doubles your odds of a reply.
Applying to Jetty? Get the contact, not the form.
The Jetty interview process
3 stages · 7 days typical · take-home: yes · modelled from similar companies
We don't yet have verified candidate reports for Jetty. What follows is the typical process for a <50-person AI company — treat it as a model, not confirmed detail.
Interview stages
Intro Call
Video call · 30 min
Culture fit and role expectations
Founder or hiring manager
Technical Deep Dive
Video call or in-person · 60 min
Past projects and problem-solving approach
Technical founder or lead
Final Round
In-person or video · 45 min
Team fit and offer discussion
Founding team
Jetty take-home assignment
Jetty includes a take-home exercise in their interview process. For AI roles, this typically involves a practical problem that takes 2-4 hours. Focus on clean, working code over premature optimization. They're evaluating how you think and communicate, not just the solution.
Jetty interview timeline
Expect days total. Compared to similar AI companies (10 days median), Jetty is faster.
Interviewed at Jetty?
Tell us how it went — stages, questions, timeline. Takes 90 seconds and it's how this page stays accurate for the next person.
Submit your Jetty interview experience →Jetty jobs, frequently asked questions
How many jobs does Jetty have open?
4 open roles at Jetty, last checked May 2026.
Does Jetty hire remotely?
Jetty doesn't have remote openings at the moment. All roles are in Montreal, Canada.
What roles is Jetty hiring for?
Jetty is hiring across Engineering, Data, Design, Operations. The most recent opening is Founding Engineer (Agent Infrastructure / Evaluation Systems).
How do I apply for a job at Jetty?
Use the apply links above, or check our guide to getting hired at Jetty.
Does Jetty respond to cold emails?
Response rate data for Jetty not yet confirmed.
Who is the hiring manager at Jetty?
At this size, hiring is usually run by a founder or department head.
How competitive is it to get hired at Jetty?
Expect 50-100 applicants in the first two weeks for AI roles at this size. Apply within 72 hours for best odds.
How many rounds is the Jetty interview?
3 stages: Intro Call, Technical Deep Dive, Final Round.
Is the Jetty interview hard?
Focus on technical depth and system design, not puzzles. Technical Interview is reportedly the most challenging round.
Does Jetty give a take-home task?
Yes, Jetty includes a take-home assignment.
How long does Jetty take to get back to you?
Around 7 days across the full process.
What should I prepare for the Jetty interview?
technical depth and system design is the priority. Show you can work autonomously — that matters more than algorithms at <50 people.
Where is Jetty based?
Jetty is headquartered in Montreal, Canada.
Get Jetty roles before they're posted
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching Jetty
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+