Introduction
For years I was fascinated by how technology evolves, but the moment I discovered Machine Learning and Artificial Intelligence, everything changed.
I wasn’t just learning to code anymore — I was learning how machines learn.
This blog is a summary of my journey into AI/ML, the concepts I’ve learned, and how these tools now help me write better code, automate tasks, and build smarter applications.
Where My Interest Started
My interest began with questions:
- How does Google Search understand meaning?
- How do self-driving cars detect objects?
- How does ChatGPT understand and generate code?
- How do recommendation engines know what I like?
These questions pushed me deeper into the fundamentals of ML and AI.
I started learning step by step — not with big models, but with foundational concepts like:
- Linear Regression
- Logistic Regression
- Decision Trees
- K-Means Clustering
- Train/Test Splits
- Overfitting and Underfitting
The moment I trained my first model that predicted something correctly… I got hooked.
What I Have Learned So Far
Over time, my skillset expanded from basic algorithms to advanced AI concepts.
### 1. Core ML Concepts
I practiced and understood:
- Supervised & Unsupervised Learning
- Classification vs Regression
- Loss functions and gradient descent
- Model evaluation (accuracy, precision, recall, F1-score)
- Feature scaling and normalization
### 2. Deep Learning
I learned how neural networks work internally:
- Input → Hidden Layers → Output
- Activation functions (ReLU, Softmax, Sigmoid)
- Backpropagation
- Batch vs Epoch
- CNNs for image tasks
- RNN/LSTM for sequence modeling
And then moved into modern frameworks:
- TensorFlow / Keras
- PyTorch
- GPU training basics
### 3. LLMs (Large Language Models)
This was the biggest turning point for me.
I studied:
- Transformers (the architecture behind GPT, BERT, LLaMA)
- Attention mechanism (“Query, Key, Value”)
- Tokenization
- Embeddings
- Prompt engineering
- Fine-tuning and LoRA
I also explored:
- Hugging Face Transformers
- Sentence Transformers
- Quantized LLMs (GGUF/GPTQ)
- Vector databases (ChromaDB, Pinecone)
### 4. MLOps Concepts
To understand how real companies use ML, I also learned:
- MLflow
- Dataset versioning (DVC)
- FastAPI model deployment
- Docker-based inference
- GPU optimization
- Model monitoring & drift concepts
How AI/ML Helps Me In Coding & Development
AI didn’t replace my coding — it supercharged it.
Here’s how:
1. Faster Development
LLMs help in:
- writing boilerplate code
- generating functions
- debugging
- converting logic from one language to another
- writing SQL queries
- designing system architecture
2. Better Problem Solving
Learning ML improved my thinking:
- I now break problems down mathematically
- I understand patterns in data
- I can analyze performance using metrics
- I think in terms of optimization
3. Automating Repetitive Tasks
I built small AI tools for:
- text analysis
- log classification
- error prediction
- auto-documenting code
- generating test cases
4. Building Smarter Products
AI gives power to normal apps:
- recommendation features
- prediction features
- chat-based support
- search engines with embeddings
- anomaly detection
Now every app I build can become “smart”.
Technical Knowledge I Gained
Here are some concepts I understand well now:
- Activation functions
- Transformers & attention
- Gradient descent
- Loss optimization
- Embeddings + vector search
- Tokenization techniques
- Fine-tuning vs prompting
- ONNX Runtime optimization
- GPU vs CPU inference
- Batch processing
- Learning rate tuning
This knowledge helps me build scalable, optimized ML pipelines.
What's Next For Me in AI/ML
My future learning goals:
✔ Training my own small LLM
✔ Implementing RAG (Retrieval Augmented Generation)
✔ Learning WebGPU for in-browser ML
✔ More work on ML + DevOps (MLOps)
✔ Building AI-powered SaaS tools
AI is growing faster than anything else in tech, and I want to be a part of that journey.
Conclusion
AI/ML completely changed the way I build software.
It made me:
- faster
- smarter
- more efficient
- more creative
And opened a world where apps don’t just run — they learn.
This journey is still going, but I’m proud of the path I’ve taken so far.
The future of development is AI-powered — and I’m excited to build in that future 🚀