ITHRON RESEARCH

Observations

A list of lessons in no particular order. Most are not original, but we can attest to their value based on our experiences over the years working with a wide range of organizations. Season with salt.

Automation

• Try quick, simple solutions first. In the worst case scenario, they can provide a baseline to assess more complex techniques.
• Very rarely can AI systems be effective drop-in replacements for human agents. The rest of the organization will have to adapt to the new system. Figuring out how to design, implement, and manage the necessary organizational changes is often more difficult than the technical challenges of building the system.
• Slow and steady automation-creep strategy (i.e. boiling the frog) only works well when the task function is continuous. When the task function contains large steep steps, automation initiatives can get stuck.
• The AI-taking-my-job risk is there but typically overblown. Most of the jobs that AI will replace will likely take the form of reduced future hires.
• Garbage in, garbage out, but only if you have a garbage-o-meter.

Management

• Paid try-outs in real world settings are much better than interviews for recruitment.
• Leet-coding is a general technical IQ test. It’s not supposed to be a reflection of the work, nor be considered in isolation. The major advantage is revealing talent signals in young people or those from disadvantaged backgrounds.
• A team that is extremely motivated will often go to extraordinary lengths to accomplish a task. Although this is usually a good thing, over time it can mask systemic inefficiencies. Periodically introducing bouts of laziness can be beneficial. There’s probably some analogy to fasting.
• Reducing complexity is the unsung skillset of unsung heroes. High value, nearly-zero visibility.
• Working from home should be an option when possible, but the best teams still work in person. In the long run, remote work ultimately exerts pressure to raise the bar for in-person work culture.
• Work-life balance is a concept that makes bad work environments more psychologically tolerable. It's much better to think about how your work contributes to and takes away from your life.

Business/Startups

• Academia is a multi-level marketing scam perpetrated by a network of de facto paper mills.
• Don’t ask a fossil fuel company what type of car you should buy (or if you should buy a car at all).
• In private, all business questions should be answered with “it depends on … ”
• The people not in the room are often just as telling as the people who are. In general, think about what’s missing from the picture. It's difficult but worthwhile.
• A business is ultimately about value capture. You may or may not create the value, but without value capture, you have no business.
• Startups should treat VC investors simply as vendors of cash (i.e. gatekeepers to capital). Once the money is in the bank, VC firms would have delivered most of their value.

Quick Summary

• We solve problems using machine learning and other data-centric technologies.
• Typically, we start with understanding the client's problem space and decide if there's an opportunity to build something useful. Then, we help them with implementation.
• Teams are recruited ad hoc from our pre-vetted talent network. We operate more like a professional association or a university alumni group. Our operating model is based on small elite teams working quickly.
• Incentives are designed around deliverables and milestones, which puts more emphasis on outcomes than billable hours.
• When the time is right, we help our clients transition operations to their existing in-house team. We also help them recruit and hire a new team if needed.
• If there's a good fit, some of our project team members may join our clients as employees to continue their work!

Guiding Principles

Small Tailored Teams

Each team is customized to the needs of a project based on skills, interests, and personal dynamics. Our recruitment method minimizes the team size and communications overhead (like low value meetings). Each team member must have critical roles, which fosters respect and team cohesion. As a result, teams move quickly, deliver better results, and avoid unnecessary costs.

Objectives and Results

We clearly define and communicate each project’s objectives and deliverables. Clarity and verification of understanding are particularly important for long-term goals, which can be used as a compass to guide team efforts. During regular assessments, if and when objectives change, the plan and the team will be reorganized accordingly. Compensations are tied to results and not on time spent, which incentivize team members to focus on the objective rather than billable hours.

Autonomy and Accountability

To enhance each team’s work experience, team members have the autonomy to determine how they work with each other. Engineers need quiet, low-distraction environments for long periods of uninterrupted work (which makes open offices very problematic for them). Others may have small children and are more productive when working remotely. More control and autonomy means team members are held to a higher level of accountability to deliver results.

Organizational Design

To support our principles, we have developed a design that enables distributed, modular operations:

Separation of concerns

Each project is placed in an independent operational container, with its own team, resources, goals, and working structures. Activities in one project have minimal impact on others, allowing for more freedom to pursue the project objectives. This system enables us to have the flexibility to organize people in different ways to best solve problems.

Project-based recruitment

We have no permanent staff. Each team member is hired specifically for a project based on an overall fit with the project needs. As a result, we can scale up and down the team as needed. This model also allows us to assess individuals frequently. Those that deliver the best work are hired more often and given more responsibilities. This environment creates a more merit-based system whereby teams are assessed based on their work rather than seniority or prior positions.

Minimal overhead

Expenses that can’t be directly tied to a project deliverable are kept to a minimum. For example, we have no central office. To be more adaptable, we have opted to use temporary workspaces and adopted remote-working whenever possible. Having flexible working locations create better experiences for team members and better work for clients and collaborators.

Talent Network

What we do

We organized ourselves as a pre-vetted network of creative technical talent that work on projects together. If you enjoy building software and data systems, we’d love to hear from you. We value people who show a willingness to learn and make the effort to help others.

We have found the model of organizing teams ad-hoc to be more flexible and effective than maintaining a permanent staff. This method allows us to keep teams small and relevant while providing more autonomy for team members to determine the best way to tackle a project.

• Build great products and services.
• Learn new skills and grow, both professionally and personally.
• Develop excellent relationships with other talented folks.

How to join

We love to hear from people of all levels of experience. There is no minimum experience required to join the network, although we will match projects that fit with your level of expertise. Don’t be discouraged if you're just starting your career. Every person was a beginner at some point, and we’d love to help you find a way to grow professionally. You may already have a full-time job, which is not a barrier to working with us, as long as your employer is aware and provides a written waiver for legal purposes.

• Review your relevant credentials (resume, LinkedIn).
• Schedule an initial interview to better understand your skill set and interests.
• Add you to our talent roster, which we’ll use to organize teams for specific projects.

A brief note about interviews
Although we find technical interviews to have some value, you are ultimately assessed by your contributions over time. Part of the advantage of our model is that there are regularly recurring opportunities to assess your abilities and work ethics. We value people who are humble and have a willingness to learn. During our interviews, we are less interested in your ability to recall minute technical details or whether you can code on a whiteboard, and more interested in how you think about problems and your interests. The goal is not to make a binary decision (to hire or not) but rather to know you better so we can match you with projects that enable you to contribute and learn. Sometimes, our first assessment is not accurate or is simply incomplete, and we take care to re-evaluate regularly to correct our oversight and expectations.

Working together

When a project is available that matches your skill set and goals, we may contact you to review the details and to get feedback. Often, we’d recommend other teammates, and you could also recommend someone we may have overlooked. You will have a say in what you work on and who you’ll be working with.

There are no permanent roles (ex. “Chief Technology Officer”). Each person is identified by the projects they are currently working on and their roles within those projects. You may fulfill multiple roles within a project.

We have no central office. Instead, we’ve opted to give the team the option to work remotely and book temporary workspaces if needed. You’ll decide as a team how, when, and where you’ll work and meet. Our teams are encouraged to work remotely most of the time, but meet in person to discuss important issues.

Compensation

Compensation is tailored to the project goals and deliverables. It is also based on the quality of work delivered as well as how well you worked with the rest of the team.

We avoid per-hour billing. Creative technical talents are not commodities, and thus cannot be treated as such. We have found it challenging to accurately determine a person’s per-hour contributions or “market rate,” which can also vary over time. Per-hour compensation also encourages micro-management, scrutinizing work down to the hour-level, which we believe to be counterproductive for creative work.

Instead, we prefer a per-deliverable compensation system, with bonuses for overall project completion. This system avoids making assumptions about a person’s expected output and examines the actual output instead. Teams also communicate better as extra efforts are directed towards regularly planning and evaluating the goals, tasks, and schedules. Benefits to team members from extra effort, such as automating workflows and reducing project risks, are more direct and immediate.

Since humans are social beings, the experience you create for your teammates has a significant impact on project success. Your ability to manage conflicts, motivate others, and show empathy to your fellow teammates are just as important as your individual contributions.

Leadership

Frank led technology development projects for research and clinical care teams at The Hospital for Sick Children (SickKids) in Toronto. He also served as the Chief Operating Officer at Spearfront Partners, an investment firm that helped startup companies grow, and was reponsible for technology due-diligence and day-to-day operations.