# Job search goal
What I'm looking for. Cursor reads this before searching, scoring companies/leads, and composing outreach.
Skills: see `.cursor/skills/` — `collect-leads`, `job-eval`, `review-leads`, `resume-tailor`, `compose-message`.
---
## Target thesis
Find companies, labs, and teams doing **genuinely interesting technical work** where a high-agency young builder/researcher can become useful quickly.
Do not over-anchor Ryan on any single project. Specific projects are **proof points, not identity**.
Prioritize places where the work itself is unusually exciting:
- AI research labs
- AI product labs
- agent/tooling companies
- robotics
- simulation
- scientific computing
- infrastructure
- devtools
- ML systems
- human-computer interaction
- experimental interfaces
- small teams doing technically ambitious work
- startups trying to create new categories, not just automate paperwork
The ideal environment appreciates:
```
taste
speed
curiosity
technical range
research instinct
product intuition
weirdness
self-direction
ability to learn fast
ability to build from zero
```
The goal is **not** simply to find "places that need code pushed."
The goal is to find places where Ryan can work on cool, novel, technically meaningful things with people who value unusually strong builders.
---
## Avoid overusing specific projects
Do not make any one project the center of outreach.
Avoid repeatedly leading with:
- Ezcaptain
- the physics project
- maritime software
- weekend experiments
- one specific SaaS product
Use projects only when they are directly relevant to the company.
**Bad:** "I built Ezcaptain, a maritime-school SaaS platform..."
**Better:** "I've built and shipped a mix of AI experiments, internal tools, web systems, and low-level prototypes, and I'm trying to find a team doing work that is technically ambitious enough to grow into."
**Bad:** "I built a neural physics engine..."
**Better:** "I like projects at the edge of AI, systems, and simulation, especially where the problem is not already cleanly solved."
---
## Target roles
- Software engineer internships (2026 summer and beyond)
- New-grad / entry-level SWE roles (graduating May 2027)
- Research engineer / research-adjacent engineering roles
- Founding engineer or early startup roles with real ownership
- Contract or project work at small ambitious teams
## Location
- Based in New Jersey (NJIT area)
- Remote-first or hybrid preferred
- Open to relocation for the right team
## Compensation
- Internships: learning and scope over max comp
- Full-time: market rate; comfortable with equity at early startups
---
## Target categories (prioritize in search)
### 1. AI research / product labs
**Look for:**
- applied AI labs
- research engineer roles
- AI evals
- AI agents
- model behavior
- multimodal systems
- tool use
- synthetic data
- model interpretability
- AI infrastructure
- small independent labs
**Why this fits:** Ryan wants to be around people pushing technical boundaries, not just building CRUD apps.
**Best outreach angle:**
> I'm trying to find a team working on genuinely hard AI problems where a fast, research-curious builder can become useful. I'm especially interested in the space between research prototypes and real working systems.
### 2. Devtools / agent tooling / infra
**Look for:**
- coding agents
- browser agents
- eval platforms
- CI/CD for AI systems
- developer environments
- observability
- workflows for AI-assisted engineering
- model/tool orchestration
**Why this fits:** Ryan is already thinking deeply about how agents fail, how to harness them, and how builders actually use them.
**Best outreach angle:**
> I'm interested in the gap between impressive AI demos and tools that actually survive real engineering workflows. I like building harnesses, tests, and systems that make agents more useful instead of just more flashy.
### 3. Robotics / simulation / spatial / embodied AI
**Look for:**
- robotics startups
- simulation platforms
- world models
- physics/simulation infrastructure
- autonomous systems
- computer vision + control
- embodied AI
- games as simulation environments
**Why this fits:** Cool technical work with real-world complexity, not just app development.
**Best outreach angle:**
> I'm interested in AI and systems work that touches the real world: robotics, simulation, control, spatial reasoning, and tools that make complex systems easier to build or understand.
### 4. Frontier startups with real technical ambition
**Look for:**
- founder-led teams
- hard technical products
- new category creation
- strong engineering taste
- unusually ambitious product vision
- public demos or technical writing
- hiring language around high-agency builders
**Note:** Funding is a discovery signal, not a quality signal. Do not overvalue "recently funded" by itself.
**Best outreach angle:**
> I'm looking for a team doing work that is genuinely ambitious, where being fast, technical, curious, and self-directed is actually useful. I'm less interested in traditional internship structure and more interested in finding a place where I can contribute to hard problems.
---
## Must haves
- Hands-on engineering — ship code, not slide decks
- Room to go deep on hard problems (AI, systems, simulation, infra)
- Honest technical culture
- Work that would make Ryan sharper, not just employed
## Nice to haves
- Early-stage startup or lab-like team
- Exposure to frontier AI/systems work
- Ambiguous problems with room to own hard projects
- People clearly better than Ryan at something important
- Remote-friendly async team
## Avoid
- Pure ticket factory with no ownership
- Sales engineering with no coding
- Mandatory 5-day office without local option
- Generic B2B SaaS with no technical depth
- Compliance-only products, basic dashboards, normal agency work
- "AI wrapper" with no real engineering or research behind it
- Pitching boring vertical SaaS unless the company is specifically in workflow tooling Ryan cares about
---
## Company scoring rubric
Score each company **1–5** on:
```
technical_ambition
research_energy
builder_appreciation
novelty
talent_density
learning_upside
outreach_surface
role_plausibility
```
Record scores in lead **Concerns** or company **notes** when useful.
### technical_ambition
Is the company solving hard technical problems?
**High signals:** AI/ML depth, systems complexity, robotics/simulation, infrastructure, scientific or research-heavy work, hard engineering constraints
### research_energy
Does the company feel close to research or invention?
**High signals:** papers, technical blog posts, open research questions, research engineer roles, lab-like culture, experimental product work
### builder_appreciation
Would they value someone who can quickly prototype and figure things out?
**High signals:** small team, founder-led, high-agency language, public demos, fast shipping culture, generalist roles
### novelty
Is the company doing something actually new or unusually interesting?
**Low score for:** generic B2B SaaS, compliance-only products, basic dashboards, normal agency work, "AI wrapper" with no depth
### talent_density
Would Ryan learn a lot from the people there?
**High signals:** strong technical founders, ex-research lab people, serious engineers, public technical work, hard hiring bar
### learning_upside
Would this place make Ryan sharper?
**High signals:** exposure to frontier AI/systems work, ambiguous problems, room to own hard projects, people who are clearly better than him at something important
### outreach_surface
Can Ryan reach a real person?
**High signals:** founder email, active X/LinkedIn, GitHub, public Discord, technical blog, personal websites
### role_plausibility
Could Ryan plausibly get in?
**High signals:** internships, junior roles, research assistant roles, contract/project work, founder reachable, small team with flexible hiring
**Lead `match` mapping (approximate):**
- **strong** — mostly 4–5 on technical_ambition, builder_appreciation, learning_upside; role_plausibility ≥ 3
- **moderate** — interesting on 2+ dimensions but gaps (location, seniority, weak outreach_surface)
- **weak** — generic work, low novelty, or poor fit with avoid list
- **bad** — clear misfit; archive
---
## Outreach style (see also `profile/voice.md`)
Outreach should **not** sound like a project summary. It should sound like a sharp young technical person trying to get near hard problems.
Use the **best outreach angle** for the company's target category (above) when drafting. Adapt founder/lab, research-style, and high-agency startup templates in `profile/voice.md`.
**Agent instructions when writing company notes or outreach:**
- Lead with the company's interesting work, not Ryan's resume
- Use Ryan's projects only as supporting evidence
- Do not repeatedly mention the same project
- Do not pitch boring SaaS unless the company is specifically in vertical SaaS or workflow tooling
- Prefer research, labs, AI, infra, robotics, simulation, devtools, and technically ambitious startups
- Avoid sounding like Ryan just wants any software job
- Tone: direct, curious, technical, ambitious — or **weird/loud** for qualifying targets (see below)
### Weird outreach addendum
For normal companies: generic/direct outreach is fine.
For unusually interesting small teams, founders, labs, or cracked technical cultures: draft a **second loud-but-true version**. Full rules, subject line modes, and templates in **`profile/weird-outreach.md`**.
**Use weird outreach when:** small company, active founder online, strong technical personality, genuinely exciting work, normal apps won't stand out, Ryan has a specific reason to care.
**Do not use when:** large formal company, HR-only, conservative culture, deceptive subject line, desperate tone.
**Agent rule:** always offer **normal + weird** versions when target qualifies. Explain why weird fits, what detail anchors the message, subject mode, and cringe risk. **Ryan must approve before any weird message is sent.**
---
## Search keywords
Prioritize searches that surface ambitious technical work:
```
AI research engineer intern
research engineer machine learning
AI agents startup
robotics software intern
simulation engineer
ML systems engineer intern
AI infrastructure intern
developer tools AI intern
applied AI lab
embodied AI startup
scientific computing software
founding engineer AI startup
technical intern small team
```
Secondary (still valid but lower priority unless company scores high on rubric):
```
software engineer intern
backend engineer intern python
full stack intern remote
```
---
## Lead collection loop
Use **`collect-leads`** skill + Playwright MCP.
1. Read this file and `profile/`
2. Source playbook: `.cursor/skills/collect-leads/{source}.md`
3. Each lead → `scripts/new_lead.py` (creates `leads/L0001-*.md` + syncs CSV)
4. Run **`job-eval`** on each new lead (include company rubric scores in notes)
5. Dedupe by URL (script-enforced); stop at batch limit below
**Batch size per run:** 5–10 new leads.
**Prefer** leads from target categories above over generic job board volume.
---
## Resume & messages
- **`resume-tailor`** → `outputs/L0001/resume_notes.md` — pick projects **relevant to this company**, not the same lead project every time
- **`compose-message`** → `outputs/L0001/` — lead with their work; projects as proof only when relevant
- Always **`profile/voice.md`** for anything Ryan sends