Day 1 of 60: Why I'm Betting Everything on This Impossible Goal
The Audacious Goal 🎯
Today, I'm starting what might be the most insane challenge of my career: transforming from a Web developer into an ML Research Engineer capable of training 70B+ parameter models and designing novel transformer architectures—all in 60 days.
The target role? ML Research Engineer at a well-funded AI company in San Francisco. They want someone who can "code up a transformer from scratch in PyTorch" and has "graduate-level ML experience." The salary range? $150K-$300K plus equity.
My current ML experience? Practically zero. No research papers, no deep learning projects, no transformer implementations. Just a solid software engineering background and a track record of learning complex technologies fast.
Why This Matters (And Why I'm Sharing Publicly) 💡
This isn't just about landing a job. It's about proving that with the right strategy, intense focus, and public accountability, you can make seemingly impossible transitions in tech.
The AI field is exploding, but there's a massive talent shortage. Companies are desperately seeking ML engineers and researchers, but most developers think they need years of academic training to make the switch. I'm betting that's wrong.
My thesis: If you can learn blockchain development, smart contracts, and DeFi protocols (which I did), you can learn the mathematics and implementation details of modern AI systems.
The Brutal Reality Check 📊
Let me be completely honest about what I'm up against:
Current Skills (Strong):
- ✅ 3+ years JavaScript/Python experience
- ✅ Full-stack development (React, Django, PostgreSQL)
- ✅ Blockchain development (Solana, smart contracts)
- ✅ Proven ability to learn complex technical concepts quickly
- ✅ Strong engineering fundamentals
Massive Gaps (Brutal Truth):
- ❌ Linear algebra at the level needed for ML research
- ❌ Deep understanding of calculus and optimization theory
- ❌ Experience with transformer architectures
- ❌ Knowledge of large-scale model training
- ❌ Understanding of cutting-edge ML research
- ❌ PyTorch expertise for implementing models from scratch
The 60-Day Master Plan 🗺️
I've designed a ruthless 60-day curriculum that assumes 12-14 hours of focused work per day:
Phase 1 (Days 1-20): Mathematical Foundations
- Linear algebra mastery (eigenvalues, SVD, matrix calculus)
- Probability theory and statistics
- Calculus and optimization theory
- Neural networks from absolute scratch
Phase 2 (Days 21-35): Transformer Architecture Mastery
- Implement transformer from scratch (no libraries)
- Master attention mechanisms and positional encoding
- Study and implement key papers (BERT, GPT, etc.)
- Advanced architectures and optimization techniques
Phase 3 (Days 36-50): Search & Embedding Specialization
- Dense retrieval and semantic search
- Embedding model fine-tuning
- Large-scale search architectures
- Custom evaluation frameworks
Phase 4 (Days 51-60): Research & Innovation
- Large-scale model training techniques
- Novel architecture exploration
- Research methodology and experimentation
- Portfolio projects that demonstrate research capability
Day 1: Starting With Linear Algebra 📐
Today's focus is on building rock-solid foundations in linear algebra. Here's what I'm tackling:
Morning (5 AM - 8 AM): Core Concepts
- Vector operations and vector spaces
- Geometric intuition behind linear transformations
- Gilbert Strang's MIT 18.06 lectures
Mid-Morning (9 AM - 12 PM): Implementation
Building a complete vector operations library from scratch:
class Vector:
def __init__(self, components):
self.components = components
def dot(self, other):
# Implementing dot product from first principles
pass
def cross(self, other):
# 3D cross product implementation
pass
def magnitude(self):
# Vector magnitude calculation
pass
Afternoon (1 PM - 4 PM): Problem Solving
- 20+ vector problems from Khan Academy
- Geometric interpretation exercises
- Linear independence proofs
Evening (5 PM - 8 PM): Advanced Topics
- Reading Chapter 1 of Strang's "Introduction to Linear Algebra"
- Creating visualization tools for vector operations
- Implementing basis transformations
The Public Accountability System 📝
To ensure I stick to this intense schedule, I'm implementing several accountability mechanisms:
- Daily Blog Posts: Every day, I'll document what I learned, what I built, and what challenged me
- GitHub Commits: All code and implementations will be publicly available
- Weekly Progress Videos: Demonstrating the concepts I've mastered
- Twitter Updates: Real-time progress sharing
- Community Engagement: Answering questions and helping others learn
Why Share This Journey? 🤝
There are several reasons I'm documenting this transformation publicly:
For Aspiring Career Changers:
If you're a software engineer wanting to break into AI/ML, this will show you exactly what's possible and what it takes.
For Current ML Engineers:
You'll see the journey from a beginner's perspective, which might help you mentor others or identify knowledge gaps in your own understanding.
For Myself:
Public accountability is powerful. Knowing that hundreds of people are watching my progress will keep me motivated during the inevitable difficult days.
The Metrics That Matter 📈
By the end of 60 days, I need to demonstrate:
Technical Capabilities:
- Can implement any transformer variant from scratch in under 4 hours
- Can explain and derive mathematical foundations of modern ML
- Can design and run large-scale training experiments
- Can read and implement cutting-edge research papers
Portfolio Evidence:
- 5+ major ML projects with clean documentation
- 3+ implementations of recent research papers
- 1+ original research contribution
- Technical blog with 60+ detailed posts
Research Mindset:
- Ability to formulate hypotheses and design experiments
- Understanding of evaluation methodologies
- Knowledge of current research frontiers
- Skill in communicating complex technical concepts
What Success Looks Like 🏆
In 60 days, I want to be able to walk into that interview and say:
"I can implement any transformer architecture from memory. I understand the mathematical foundations deeply enough to derive backpropagation equations. I've trained models on multi-GPU setups and designed novel architectures. Here's my GitHub with 20+ implementations from scratch, and here's my blog documenting every step of the journey."
The Reality Check (Again) ⚠️
Let me be crystal clear: this might fail. 60 days to go from web developer to ML researcher is genuinely insane. Most people spend years in PhD programs learning what I'm trying to master in 2 months.
But here's what I know:
- I've successfully made rapid transitions before (web development → blockchain development)
- I have strong mathematical aptitude (engineering background)
- I'm willing to work 14-hour days for 60 straight days
- I have a systematic approach and clear milestones
Join Me on This Journey 🚀
Whether you're:
- A developer considering a career change to AI/ML
- An ML engineer curious about the learning journey
- Someone who loves seeing impossible goals attempted
- Just interested in the intersection of education and intensity
I invite you to follow along. I'll be sharing:
- Daily progress updates and lessons learned
- All code implementations and mathematical derivations
- Weekly deep-dives into complex topics
- Real-time problem-solving and debugging
- The emotional journey of such an intense learning experience
Follow my progress:
- 📝 Daily blog posts here on dev.to
- 💻 Code implementations on GitHub
- 🐦 Real-time updates on Twitter @VivekJami4
- 💼 Professional updates on LinkedIn
Day 1 Commitment 💪
As I write this, it's 11:30 AM. I'm about to start my first 14-hour learning day. My coffee is ready, my notebooks are open, and Gilbert Strang's linear algebra lectures are queued up.
The journey from Web3 developer to ML Research Engineer starts now.
Will I make it? I honestly don't know. But I'm going to document every step, every breakthrough, and every failure along the way.
See you tomorrow for Day 2: Matrix Operations and the Path to Understanding Neural Networks.
What do you think? Is this goal realistic or completely insane? Have you made similar career transitions? Drop a comment below—I'd love to hear your thoughts and experiences!
Tags: #MachineLearning #CareerChange #AI #DeepLearning #60DayChallenge #TechTransition #LinearAlgebra #PyTorch #Transformers #MLResearch
P.S. If you're attempting something similarly ambitious, I'd love to connect. Sometimes the craziest goals need the craziest people to attempt them together.
Top comments (0)