Team Project
also known as collaborative project, group project, team-based learning, TBL project, cooperative project
A group of learners tackles a shared project challenge together, dividing roles and responsibilities while building toward a single collective output. The team project pattern is distinct from individual project work in one critical respect: the interpersonal and collaborative competencies — communication, role negotiation, conflict resolution, shared accountability — are themselves learning targets, not just preconditions. Larry Michaelsen formalised Team-Based Learning at the University of Oklahoma in the 1970s. He established that most reported 'problems' with learning groups arise from inappropriate group formation and assignment design, not from collaboration itself. In AI and tech upskilling, team projects are the dominant format for professional preparation: nearly all production AI work is team work. The team project pattern teaches learners not just what to build but how to build it together.
How the learner advances
Intent. Build both domain competence and collaborative work skill by making a group of learners jointly accountable for a shared product — so that neither competency can be acquired without the other.
When to apply. Apply when two conditions hold simultaneously: the target domain requires collaborative practice (which AI development always does), and the learners need to develop collaborative working skill explicitly, not assume it. Team projects are the right format when the team will face division of labour, coordination overhead, and genuine interdependency — where one member's output is another's input. Do not use team projects when the goal is purely individual skill assessment, or when the team size is too large (beyond five to seven) for every member to be genuinely accountable.
Threshold — earns the next step. Each team member can describe specifically what they contributed and what depended on that contribution, and can name at least one collaborative practice they would apply differently in the next team project.
Masterpiece — the artifact that proves it. The shared team product — working system, deployed tool, research deliverable — that no single team member could have produced alone, accompanied by each member's individual reflection on their collaborative contribution and growth.
Facets
- Container — async
- Mode — collaborativeapplied
- Reach — team
- Persona — developeranalyst-opsnon-technicalmanager-leader
- Craft (AI Fluency) — collaborationsynthesisdiscernment
- Learner — human
- Trainer — human
Inputs
- Sufficiently complex shared challenge — A project scope large enough that no single individual could reasonably complete it in the time available — creating genuine interdependency. If a project is completable solo, team structure produces free-riding rather than collaboration.
- Deliberately formed team (3–5 learners) — Teams formed by the instructor with explicit attention to complementary skills and diversity of background — not self-selected friendship groups. Michaelsen's research shows that self-selected groups underperform instructor-formed diverse teams on both learning and product quality.
- Individual accountability mechanisms — Assessment structures that make each member's individual contribution visible and evaluated — peer assessment, individual readiness tests, role logs, or code attribution. Without these, free-riding is rational.
- Team charter and shared norms — A brief written agreement negotiated by the team at the outset: decision-making norms, conflict escalation path, communication cadence, and definition of acceptable contribution. Teams that skip this step spend conflict budget on meta-disputes rather than project work.
Outputs
- A more capable learner — A learner who has practised not just domain skills but the collaborative moves professional AI work requires: asynchronous coordination, technical communication across skill levels, disagreement-to-decision, and shared ownership of a shipped product.
- Shared team product — The collective artefact — the masterpiece — built by the team and attributable to the team's joint effort: a working system, a research deliverable, a deployed tool. The product's quality reflects both domain competence and collaborative effectiveness.
Steps (5)
Form teams and negotiate charters
The instructor forms teams with attention to complementary skills and diverse perspectives. Each team negotiates and signs a team charter covering norms, roles, and conflict resolution before project work begins.
Scope and plan together
The team breaks the project challenge into components and assigns ownership — not just tasks. Ownership means the assignee is accountable for the component's quality and its integration with adjacent work. The team then builds a shared timeline with explicit dependency mapping.
Parallel work with integration checkpoints
Team members work on their components simultaneously, but the team holds regular integration checkpoints — not just status updates — where components are actually assembled and tested together. Integration failures surface early, not at the deadline.
Peer review and collective revision
Before the final deadline, team members review each other's components for quality and integration fit. The team collectively owns the revision — not just the original assignee. This is the moment where the team's collaborative norm is most tested.
Present and retrospect
The team presents the finished product collectively. After presentation, a structured retrospect surfaces what collaborative practices worked, what created friction, and what each member would do differently. The retrospect targets collaborative competency explicitly, not just technical lessons.
Principles
- Interdependency, not division — a project split into independent pieces where each member does their section alone is parallel individual work, not a team project. Genuine team projects require integration points where one member's output becomes another's input.
- Individual accountability within collective ownership — free-riding collapses team projects; the accountability structure must make each member's contribution visible and consequential.
Known uses (4)
Team-Based Learning — Michaelsen (University of Oklahoma, 1970s onward) — University of Oklahoma / Team-Based Learning Collaborative
higher education and professional training Larry Michaelsen formalised TBL at the University of Oklahoma in the 1970s, codifying the conditions under which groups become high-performing learning teams. The framework has been adopted in medical, legal, business, and engineering educ…
Stanford CS229 / CS230 team projects — Stanford University
university AI education Stanford's flagship ML and deep learning courses both use team projects as the primary assessment vehicle, with teams of two to three presenting to panels of instructors and TAs. The team project format mirrors the collaborative structure…
MLH Global Hack Week team format — Major League Hacking
developer education and hackathons MLH's non-competitive Global Hack Week explicitly structures learning in teams to develop both technical and collaborative skills, with workshops and builds done in cross-disciplinary groups of two to five.
Gauntlet AI Catalyst — team build cohort — Gauntlet AI
enterprise / workforce AI upskilling An AI-upskilling cohort in which a team works together on a real capstone drawn from the employer's own roadmap — so the collaborative output is an actual business deliverable with genuine integration dependencies, not a course exercise.
Known failure modes (3)
- [parallel-individual-work]
The anti-pattern of a 'team project' that is actually a collection of independent individual pieces assembled at the end. Without genuine integration dependencies — where one member's output gates another's work — there is no collaborative learning, only coordination overhead.
- [free-rider-tolerance]
The anti-pattern of team assessment structures that allow a non-contributing member to receive the same grade as contributors. When individual accountability is absent, rational learners reduce their own effort to match the minimum required. Peer assessment and role logs are the standard mitigations.
- [friendship-group-formation]
The anti-pattern of allowing learners to self-select into friend groups. Michaelsen's research consistently shows that instructor-formed diverse teams outperform self-selected homogeneous groups on both learning outcomes and product quality. Self-selection optimises for comfort, not growth.
Related trainings (4)
- Project-Based Learning★★
Build deep, transferable knowledge by making learners the investigators of a genuine question or problem, not the recipients of pre-packaged answers.
- Hackathon★★
Demonstrate and develop the ability to ship a functional artefact under real time pressure and constraint — replacing theoretical competence with demonstrated delivery capability.
- Capstone Project★★
Require the learner to integrate and apply everything learned across a programme into one substantial, publicly defensible piece of work — proving readiness to practise.
- Problem-Based Learning★★
Force learners to build the knowledge they need by confronting an ill-structured real problem before they have the answers — making the acquisition of content purposeful rather than preparatory.
Sources (3)
Michaelsen, L.K., Bauman Knight, A., & Fink, L.D. (2004). Team-Based Learning: A Transformative Use of Small Groups in College Teaching. Stylus Publishing.
“Most of the reported 'problems' with learning groups (free-riders, member conflict, etc.) are the direct result of inappropriate group assignments.”
Team-Based Learning — Wikipedia
Catalyst — Gauntlet AI
“They work together to accelerate your roadmap by bringing a real capstone project to life.”
Provenance
- Ecosystem: education
- Added to catalog:
- Last updated:
- Verification status: verified