30-60-90 Day Plan Template for Founding Engineer at Seed‑Stage AI Startup

The candidates who prepare the most often perform the worst. In the June 12 2024 hiring loop for a founding engineer at Aurora AI, the senior PM whispered, “He spent the whole screen time on TensorFlow ops latency without ever naming the product’s 8‑week data‑pipeline sprint.” The loop vote was 4‑1 against hire. The following war‑stories prove why a razor‑sharp 30‑60‑90 plan beats generic ambition.

What should a founding engineer prioritize in the first 30 days?

Details for this section: Aurora AI (seed‑stage video‑summarization), interview question “Design a data‑pipeline for real‑time summarization,” debrief vote 3‑2 for hire, compensation $185,000 base + 0.04 % equity, framework “Google 4‑Quadrant Impact Matrix,” candidate quote “I’d start by scaling the model,” date June 15 2024.

First 30 days demand execution of a single end‑to‑end data‑pipeline prototype that reduces latency from 1.8 seconds to sub‑500 ms on the Aurora AI video‑summarization service. The judgment: ignore breadth, double‑down on a measurable latency reduction that the hiring manager, Maya Lee (Director of Engineering), flagged as the deal‑breaker.

Maya Lee emailed the candidate on June 15 2024:

> Subject: 30‑60‑90 Plan – Week 1 Focus – Aurora AI

> “Your first deliverable is a working pipeline that ingests 10 K videos/day, runs inference under 500 ms, and logs latency in CloudWatch. No UI, no docs, just the pipeline.”

During the loop, the senior engineer, Raj Patel, asked, “How would you handle schema drift in the incoming video metadata?” The candidate answered, “I’d add a versioned protobuf.” The debrief note: “He solved the latency but ignored schema drift—a red flag for long‑term maintainability.” The loop voted 3‑2 for hire because the latency win outweighed the schema oversight.

The 30‑day judgment: deliver a latency‑focused prototype, document results in a one‑page PR/FAQ, and schedule a post‑mortem with the data‑science lead, Priya Kumar (Senior ML Engineer).

How does the 60‑day roadmap differ for a seed‑stage AI startup?

Details for this section: Aurora AI’s Series A closing on August 1 2024, product milestone “offline‑mode support by day 60,” interview question “Explain your trade‑off between model size and inference speed,” debrief vote 4‑1 for hire, compensation $185,000 base + 0.04 % equity, framework “Amazon Mechanism Design Checklist,” candidate quote “I’ll prune the model,” date July 3 2024.

By day 60 the founding engineer must extend the prototype into a production‑ready microservice that supports offline inference for edge devices, a requirement highlighted by the CTO, Evan Shapiro, during the July 3 2024 debrief. The judgment: shift from latency‑only to durability‑and‑offline capability, because seed investors care about market‑ready features, not isolated benchmarks.

Evan Shapiro told the candidate, “We need an offline fallback that runs on a Raspberry Pi 4 with 2 GB RAM and still delivers < 2 seconds summary.” The candidate replied, “I’ll prune the model to 80 % of parameters.” The loop noted that the candidate’s answer aligned with the “Amazon Mechanism Design Checklist” item 3, “Validate edge constraints before scaling.”

The 60‑day plan includes: (1) building a Docker image under 250 MB, (2) adding a fallback inference path that uses ONNX Runtime, (3) writing integration tests that simulate network loss, and (4) presenting a performance report to the board on August 5 2024. The board’s approval vote was 4‑0 in favor of the candidate once the offline demo proved sub‑2‑second latency.

The 60‑day judgment: broaden the scope to offline reliability, embed edge constraints early, and lock in board confidence with a concrete demo.

> 📖 Related: Fudan Pm Courses Guide 2026

Which metrics prove a founding engineer succeeded by day 90?

Details for this section: Aurora AI’s KPI “monthly active users (MAU) × average summary length,” metric target 15 % MAU growth, interview question “What success metrics would you track after launch?” debrief vote 5‑0 for hire, compensation $185,000 base + 0.04 % equity, framework “Stripe Payments Impact Rubric,” candidate quote “I’ll track latency and crash rate,” date August 20 2024.

By day 90 the engineer must demonstrate impact on Aurora AI’s core KPI: a 15 % lift in MAU × average summary length, measured via the internal analytics dashboard built on Snowflake. The judgment: success is defined by a composite metric, not by raw latency numbers; the board will reject a candidate who only shows a 30 % latency gain without business lift.

During the August 20 2024 debrief, the VP of Product, Lina Gomez, asked, “What success metrics would you track after launch?” The candidate answered, “I’ll track latency and crash rate.” The debrief note: “He missed the business‑level KPI—critical for seed‑stage funding.” The loop upgraded the vote to 5‑0 for hire after the candidate added a KPI‑driven goal in the revised 30‑60‑90 document.

The metric proof includes: (1) a 12 % reduction in average latency, (2) a crash‑rate below 0.2 %, and (3) a 17 % increase in the composite MAU × summary metric. The board’s final sign‑off on September 1 2024 required all three numbers to be present.

The 90‑day judgment: tie technical wins to business KPIs, present a triad of latency, stability, and MAU impact, and secure board sign‑off with a concise slide deck.

What internal signals do hiring committees look for in a 30‑60‑90 plan?

Details for this section: Aurora AI HC (Hiring Committee) on September 5 2024, internal rubric “Founding Engineer Impact Score (FEIS) ≥ 8,” vote count 4‑1 for hire, compensation $185,000 base + 0.04 % equity, framework “Google 4‑Quadrant Impact Matrix,” candidate quote “My plan aligns with the FEIS,” date September 2 2024.

Hiring committees score the plan against the FEIS rubric, which demands ≥ 8 on four dimensions: technical depth, product impact, team leadership, and risk mitigation. The judgment: a plan that scores 9 on depth but 4 on risk mitigation will be rejected, because seed investors prioritize risk awareness.

On September 2 2024 the candidate emailed the committee:

> “My 30‑60‑90 plan maps each deliverable to the FEIS rubric, delivering a 9‑point technical depth score and a 7‑point risk mitigation score.”

The HC chair, Carlos Mendoza, responded, “Raise risk mitigation to ≥ 8 or we cannot proceed.” The committee vote on September 5 2024 was 4‑1 for hire after the candidate added a contingency plan for model drift, boosting risk mitigation to 8.

The internal signal: a revised plan that explicitly references the “Google 4‑Quadrant Impact Matrix” and includes a risk‑mitigation appendix convinces the HC. The judgment: embed the committee’s rubric language verbatim and exceed the minimum on every dimension.

> 📖 Related: PM Role Transition from IC to Manager at Amazon: Year 1 Survival Tips

How to align compensation expectations with a 30‑60‑90 plan?

Details for this section: Aurora AI offer letter dated September 10 2024, base $185,000, equity 0.04 % vesting over 4 years, sign‑on $35,000, interview question “What salary range do you expect?” candidate quote “I target $190k base,” debrief vote 5‑0 for hire, framework “Amazon Total Compensation Calculator,” date September 8 2024.

Compensation aligns when the 30‑60‑90 plan delivers the KPI boost that justifies the $190k‑base ask. The judgment: tie each compensation component to a concrete deliverable; otherwise the candidate appears “price‑driven,” and the board vetoes the offer.

During the September 8 2024 interview, the recruiter, Jenna Kim, asked, “What salary range do you expect?” The candidate replied, “I target $190k base.” The debrief note: “His ask matches the market for a founding engineer with $185k base and 0.04 % equity, but we need a deliverable‑linked justification.”

The offer letter on September 10 2024 included a clause: “Equity vests upon achieving the 90‑day KPI of 15 % MAU lift.” The board’s final sign‑off required that the candidate’s 30‑60‑90 plan explicitly references this clause. The judgment: embed compensation triggers in the plan to secure the full package.

Preparation Checklist

  • Review Aurora AI’s public blog on the video‑summarization product released March 2024; note the 8‑week roadmap.
  • Draft a one‑page PR/FAQ that maps each deliverable to the “Google 4‑Quadrant Impact Matrix.”
  • Run a latency benchmark on a 10 K‑video batch using TensorFlow 2.12 on an n1‑standard‑4 GCP instance; record sub‑500 ms results.
  • Build a Docker image under 250 MB and push to Aurora AI’s private GCR registry; verify with “docker image inspect.”
  • Work through a structured preparation system (the PM Interview Playbook covers “Edge‑Constraint Trade‑offs” with real debrief examples).
  • Prepare a risk‑mitigation appendix that cites the “Amazon Mechanism Design Checklist” item 3.
  • rehearse the “FEIS rubric alignment” paragraph for the September 5 2024 HC meeting.

Mistakes to Avoid

BAD: “Focus on UI polish in week 1.”

GOOD: “Deliver a latency‑focused pipeline; UI comes after day 30.” The problem isn’t polishing the UI — it’s mis‑allocating scarce engineering weeks.

BAD: “Quote a salary range without linking it to a KPI.”

GOOD: “State $190k base and tie equity vesting to a 15 % MAU lift.” The problem isn’t the salary number — it’s the lack of deliverable‑based justification.

BAD: “Ignore the FEIS risk‑mitigation score.”

GOOD: “Add a model‑drift contingency plan to raise risk mitigation from 4 to 8.” The problem isn’t the score itself — it’s the missing mitigation strategy.

FAQ

Does a 30‑60‑90 plan guarantee a hiring decision? No. The decision hinges on the FEIS score; a plan that hits all rubric points and ties compensation to KPI will sway a 4‑1 vote, but missing any dimension can cause a reject.

Can I reuse a generic template from a non‑AI startup? No. Seed‑stage AI startups like Aurora AI demand product‑specific latency and offline constraints; a generic template will fail the “Edge‑Constraint Trade‑off” test and be vetoed by the HC.

What if my salary expectation exceeds the offer? Not the base figure — the equity trigger is the lever. Align your 90‑day KPI to the equity clause; the board will approve a higher base only if the KPI promise is credible.amazon.com/dp/B0GWWJQ2S3).

Related Reading

What should a founding engineer prioritize in the first 30 days?