ML & Hackathon Competition Career — Master Roadmap

Phase 1: Foundation (Months 1–2)

Goal: Set up tools, learn the formats, get your first submission on the board.

• Set up Python stack: numpy, pandas, sklearn, xgboost, lightgbm, pytorch
• Complete Kaggle Learn micro-courses (free): Python, Pandas, ML Intro, ML Intermediate, Feature Engineering, Deep Learning
• Solve 3–5 past Kaggle competitions (just practice, no submission)
• Join first Kaggle "Getting Started" competition (Titanic, Housing Prices)
• Attend your first MLH or Devpost online hackathon (learn the format — winning is secondary)
• Create accounts: Kaggle, Devpost, DoraHacks, WorldQuant BRAIN
• Start solving Jane Street monthly puzzle (janestreet.com/puzzles) — habit from day 1
• Watch: NVIDIA GTC 2026 free virtual keynote (free, March)

Phase 2: Building Momentum (Months 3–5)

Goal: Build a streak, find your niche, get first wins.

• Enter 1 active Kaggle competition per month — aim for top 50%
• Enter 1 hackathon per month from Devpost, MLH, or DoraHacks
• Build your GitHub portfolio repo: 1 clean notebook per competition
• Read ALL winning solutions (Kaggle Discussion tab) for competitions you join
• Start Zindi or DrivenData competition (fewer competitors = realistic first win)
• Enter WorldQuant IQC 2026 (free, runs Mar–Sep, $100K prize pool)
• Attend Google Cloud Next or Databricks Summit (free online) — learn production ML
• Subscribe to: mlcontests.com newsletter + dev.to/dorahacks monthly digest

Phase 3: Competing Seriously (Months 6–12)

Goal: Get medals, win prizes, be seen by recruiters.

• Target top 25% on Kaggle → earn Silver/Gold medals → reach Expert rank
• Enter specialized competitions: Solafune (geospatial), AIcrowd, Codabench
• Enter IMC Prosperity next cycle (Python trading bots, $50K prize)
• Enter Citadel Data Open datathon — apply via Correlation One
• Target DoraHacks hackathons with $15K–$50K prize pools
• Build a team — find teammates on Kaggle forums, Discord, LinkedIn
• Enter 1 DoraHacks or ETHGlobal Web3 hackathon (learn new domain + bigger prizes)
• Document every competition in 04_COMPLETED with lessons learned
• Attend 1 in-person event: ETHDenver, local hackathon, or university ML club event

Phase 4: Recognition & Career (Year 2+)

Goal: Convert wins into career capital.

• Reach Kaggle Expert or Master tier
• Target ARC Prize or NeurIPS competition track (prestige tier)
• Publish writeup on Medium / Towards Data Science / your own blog for each win
• Apply for full-time ML roles — attach competition wins to every application
• Attend NeurIPS or ICML (in-person or virtual) — student scholarship available
• Apply for Jane Street FTTP or AMP program
• Mentor newcomers on Kaggle / Zindi — builds network and reputation fast
• Apply for WorldQuant BRAIN Research Consultant (if IQC performance is strong)

---

The Quant Track (Parallel Path — Start Month 2)

Run this alongside your ML competition track. Different skill set, elite ceiling.

• Month 2: Create WorldQuant BRAIN account → understand how alphas work → submit 5 test alphas
• Month 3: Enter IQC 2026 seriously — target Stage 2 qualification
• Month 4–5: Learn market microstructure basics (order books, bid/ask, spread)
• Month 6: Enter IMC Prosperity
• Month 8+: Apply to Citadel Datathon, Jane Street FTTP
• Ongoing: Solve Jane Street monthly puzzles every month without fail

---

Key Mindset Rules

1. One competition at a time. Don't spread thin across 5 at once.
2. Always read the top solutions. Even if you didn't win.
3. Ship working > ship perfect. Especially in hackathons.
4. Team up early. Solo is hard. Diverse teams (ML + design + pitch) win hackathons.
5. Presentation wins hackathons. Models win ML competitions. Know which you're in.
6. Events are not optional. Attend free virtual events every quarter minimum.
7. Document everything. Your 04_COMPLETED log IS your portfolio.
8. The quant track is harder but higher ceiling. Both tracks compound your career.