Topcoder

Forward-deployed engineer might be the most in-demand role in AI right now.

It's also one of the hardest to hire for.

Because it takes a rare mix of skills that don't usually live in the same person.

A few that matter most today:
→ Agentic AI architecture
→ Comfort with ambiguity
→ Translating business problems into technical ones

That's 3 of 10 - and the other 7 are the ones most engineers overlook. Full list below👇

New challenge: build an AI Deal Scoping Assistant.

You know the grind. Customer requirements land as a mess - RFPs, discovery notes, emails, half-structured docs - and turning that into a real solution proposal takes an expert days: a PRD, a cloud architecture, a data and integration plan, an AI approach, and an effort and cost estimate.

So build the tool that does the first draft. Drop in the requirements, and out comes a structured, traceable package - PRD, functional scope, cloud-specific architecture, data and integration strategy, AI approach, and a delivery estimate. Every output traces back to the requirement behind it, and change one requirement and it shows you what's affected. AI does the heavy lifting; a human still makes the call.

A few things worth knowing:
→ Counts toward the AI Exponential League
→ Any stack, no paid AI needed

Register and start building - link in the comments.

We're hiring a Forward Deployed Engineer for Gemini CX.
Pay: $3,000 per week.

Designing, building, and deploying agentic customer experience solutions for enterprise clients. This isn't chatbot configuration. You'll be:
→ Architecting multi-agent orchestration and conversational workflows
→ Building secure integrations into enterprise and legacy systems
→ Standing up telemetry and monitoring for production rollouts
→ Working directly with enterprise teams - no three layers of handoff

25 weeks. Full time. Must be based in the United States (remote work).

If you've shipped production-grade agentic AI and know Dialogflow CX or the wider Google CX stack, this one was scoped with you in mind.

Apply below - link in the comments.

Meet Jaco Cronje - one of the people who makes Marathon Match what it is.

Jaco competed in the very first Marathon Match back in the early 2000s. Two decades later, he's mostly on the other side now: testing problems before they ever reach you. He writes his own solutions, hunts down the seed that crashes the tester, finds the edge case that breaks the scoring, and makes sure every problem is easy to start but genuinely hard to master.

It's careful, mostly invisible work - the kind you only notice when it's missing.

And lately it's gotten more interesting. As AI gets better at solving the straightforward stuff, Jaco's focus has shifted to designing problems where human creativity, intuition, and a clever idea can still beat the models. As he puts it, that's the part worth protecting - the reason these challenges stay fun.

When he's not breaking problems on purpose, he's out hiking, running, or growing chillies in his greenhouse.

Twenty years in, still making the challenges better - one broken edge case at a time.

Thanks for everything, Jaco.

Check out the full interview in the comments.

Open engagement: Informatica Data Integration Developer.

This one is a technical lead role on Informatica Data Management Cloud (CDI, CAI, CDQ) - designing and architecting scalable data integration platforms, optimizing performance, and enforcing best practices across the team.

You'd own the technical design of integration solutions, build CI/CD pipelines for integration workflows, lead design reviews and code reviews, and diagnose and resolve performance bottlenecks across data integration jobs - collaborating with cross-functional teams throughout.

What's needed: proven CDI, CAI, and CDQ development expertise, strong API development and RESTful services design, advanced Spring Boot and Java, and a track record of resolving performance issues in integration workflows.

Remote, full time, $650/week, immediate start.

Link in the comments.

A single factory scan can take an engineer weeks to turn into a usable 3D model.

We're launching a challenge to do it with AI in a fraction of the time.

Here's the problem. When you 3D-scan a real factory you get millions of raw points - an exact snapshot of every staircase, handrail, conveyor and beam, but a cloud of dots a computer can't actually read. Today an engineer turns that scan into a model by hand, identifying and rebuilding every object one at a time. Thousands of them per building.

The challenge: build the AI that does it automatically - read the messy scan, recognize each object, and generate a clean, editable 3D model an engineer could actually work from.

It's a real computer-vision problem on real industrial data, not a toy dataset. Precise, messy, and hard in the way the interesting problems always are.

If you work in computer vision or 3D - or just want to see what this kind of challenge looks like - it's worth a look.

$4,300 in total prizes.

Link in the comments.

New Open Engagement: Oracle EPM Functional Consultant (India Only)

We’re seeking a senior Oracle EPM Functional Consultant to support post-migration validation across Oracle Planning (PBCS / EPBCS) environments.

This role is strictly focused on validation and testing - ensuring the target EPM system accurately reflects the source environment after migration. No migration ex*****on or defect-fixing responsibilities.

You will work closely with QA and business teams to validate:
• Data integrity & reconciliation
• Smart View reports, forms & dashboards
• Security, dimensions & metadata
• Business rules, calculations & KPIs
• UAT support, test ex*****on & reporting

Required:
- 6+ years Oracle EPM Cloud experience (Planning / EPBCS), strong Smart View, testing/UAT and functional knowledge.

Remote work. 4 weeks. Full time.

Link in the comments.

New AI Exponential League challenge: build an AI Workflow Puzzle Builder in 48 hours.

Real AI apps aren't a single prompt - they're multi-step workflows that fail in messy ways. A model returns bad output, a tool times out, a step gets invalid input.

This challenge is to build a browser-based game where players assemble AI workflows, break them with controlled failures, and repair them:
→ Validator rejects bad output - add a retry or repair step
→ A tool times out - switch to a fallback
→ A risky action needs sign-off - route it to human review
→ A workflow breaks mid-run - resume from the last good step

The goal isn't a workflow that runs. It's one that survives when something goes wrong.

Open to all levels, beginner to advanced. 48 hours on the clock.

Link in the comments.

2 days left to join our exploratory testing challenge for a major mobile banking app.

Real accounts. Real devices. Real bugs - testing the way actual customers use the app, not just test scripts.

Know someone with an eligible account? Refer them and earn $25 for every valid bug they find - no cap.

User flows to test:
✅ Onboarding
✅ Dashboards and transfers
✅ Bill pay flows
✅ Navigation and error handling

Top prize $500, payouts down to 12th place.

Challenge runs through August 31, 8:30 AM EDT.

Link in comments.

Earn $25 for every person you bring to our mobile banking app testing challenge who finds a valid bug.

No cap - every qualifying referral pays. To credit you, the person you refer adds a short Referral.txt file naming you in their submission.

This challenge is exploratory testing for a major mobile banking app, tested the way a real customer would - onboarding, dashboards, transfers, bill pay, navigation, error handling. Real accounts, real devices, real bugs.

→ Top prize: $500, payouts down to 12th place
→ Challenge runs through August 31.

Link in the comments.