Abhishek
@axabhishek_0309
Followers
597
Following
2K
Media
159
Statuses
1K
CSE Undergrad | Full Stack & AI Developer | web3 | ML , NLP & Scalable Systems | Sarcasm
Joined January 2024
Glad to share that my patent on AI Based Cooling with PCM Immersion for Data Centers is now published!⌛️ . It integrates AI, digital twins & phase change tech to boost thermal efficiency.📃. 🔗 Application No: 202511056772 A.📅 Published: 27/06/2025 . #Patent #AI #DataCenter #PCM
10
0
43
Building in public is fun until: . 🔸You push the code, and its invisible. 🔸Tweet counts lesser impressions than your counted bugs in the codebase. 🔸You start doubting the whole idea from zero. Then you push again, .not desperately .Coz thats what builders do.
0
0
2
August is more about relationship leaves than working days💀 . Aug 1: Girlfriend's Day .Aug 3: Boyfriend's Day .Aug 9: Boy Bestfriend Day .Aug 10: Girl Bestfriend Day .Aug 15: Ex Day .Aug 18: Couple Day .Aug 20: Crush Day .Aug 30: Breakup Day . whoever is having a day booked on.
0
0
3
You don’t need a PhD to become an AI Engineer. But you do need: .→ consistent effort.→ curiosity.→ patience to debug when nothing works 😮💨😮💨 . Start from Python. End up fine tuning LLMs.
0
0
2
Step 7: Research + Open Source (Ongoing) . → Read 1 paper/week (arXiv sanity, PapersWithCode).→ Reproduce models (BERT, LLaMA, DINO).→ Contribute to Hugging Face / LangChain / OSS.
1
0
2
Step 6: MLOps (Optional but 🚀lit) . → Versioning: Git, DVC.→ Tracking: MLflow, WandB.→ Docker + CI/CD . Learn how to deploy and monitor models properly.
2
0
2
Step 5: Deep NLP / Transformers (1 month) (if you wanna go in depth) . → Attention mechanism.→ BERT, GPT, T5, RoBERTa.→ Hugging Face Transformers + PEFT . 💻 Build: Chatbot, QA system, summarizer.
1
0
2
Step 4: Core NLP (1 month) . → Tokenization, BoW, TF IDF, Word2Vec.→ Libraries: nltk, spacy, gensim . 💻 Build: Sentiment analyzer, text classifier.
1
0
2
Step 3: Deep Learning (1 month) . → MLPs, CNNs, optimizers (Adam, SGD).→ Loss functions, regularization, dropout.→ Build with PyTorch (or Keras for ease) . 💻 Projects: MNIST digit recognizer, cat dog classifier.
1
0
2
Step 2: Classical ML (10 15 days) . → Regression, SVM, Trees, K-means.→ Overfitting, model metrics (accuracy, F1, ROC).→ Library: scikit learn . 💻 Project: Iris classifier, spam detector.
1
0
2
Step 1: Learn Python (Get an Idea) . → Basics: loops, functions, classes.→ Libraries: NumPy, Pandas, Matplotlib . Build small projects: calculator, data visualizer.
1
0
2
🧵 Post . Want to become an AI Engineer with deep focus on Deep Learning + NLP but don’t know where to start?. Here’a clean roadmap to move with:.
1
0
7
If your August isnt about: .Trees,Graphs,dp covers .Ai product builds .Scalable backend setups .A little gm over sol . Tough to say, ngmi.
2
0
12
P2099:🎯 .Find Subsequence of Length K With the Largest Sum.You're given an array nums and an integer k. Find a subsequence of length k that has the largest possible sum, keeping original order intact. 🧠Approach:.1.) Paired each element with its index → (value, index). 2.)
0
0
4
SDE life is just about .Opening PRs .Getting 7/8 comments .Fixing 5/6 of them .Arguing over rest 1/2 .Merging after 3/4 days . Then realization hits after you brake the prod after 10 mins.
1
0
5
Web Dev mein naya hu, .HTML mein try catch ka syntax kya hota hai, if anyone knows pls help.
0
0
2
ML models rarely fail in training, while they silently fail in prod. Reasons generally are .Pipeline Drift sometimes .Broken Preprocessing .Disabled Logging .Monitoring these factors is mandatory part.
1
0
10