Uptiq Interview Experience – R&D Role (Internship + PPO)

Company: Uptiq Roles Offered: R&D, QA, SDE, DevOps My Role: R&D Timeline: End of September

📋 Eligibility Criteria

Criteria Requirement CGPA 6.74 and above (most shortlisted were ~9 CGPA) 10th & 12th 60% and above Internship Duration 6 to 12 months Stipend INR 22,000 – 25,000/month CTC INR 22.5 LPA

Round 1: Online Assessment (OA)

Shortlisted for OA: 474 students
Duration: 2 hours, starting 8:00 AM

DSA (2 Questions):

Kadane's Algorithm
Dutch National Flag
Course Schedule
A DP-on-strings problem
A prefix-sum/suffix-sum based question

Overall difficulty: standard LeetCode Medium level

MCQs (84 total):

CS Core: OS, DBMS, SQL, SQL debugging, CN, OOP
Programming/Web: C, HTML, CSS, JS, React

✅ Result: 77 candidates shortlisted for interviews. (Company mentioned offers would be a mix of Internship + PPO and Internship + Performance-based PPO.)

Round 2: Technical Interview (L1)

Time: 5:30 PM (one of the last slots of the day)
Duration: ~30 minutes
Interviewer: A PICT alumna

Topics Covered:

Self-introduction
Deep dive into projects using LangChain, LangGraph, and CrewAI — she was impressed I had used CrewAI alongside a CI/CD pipeline
She was glad to learn I had interned at FluffyNTech, a company founded by a PICT alumnus (possibly her batchmate)
Detailed discussion of previous internship work
RAG and Agentic RAG — working, architecture, and design choices
Vector databases, FAISS, vector search, and comparison algorithms — in depth
FastAPI and Docker basics
She explained what Uptiq is building — their product Document.ai — which aligned closely with my project and internship background

✅ Result: ~30–35 candidates shortlisted for the next round.

Round 3: Technical Interview (L2)

Panel: Director + HR (conducted together)
Duration: ~40 minutes

Technical Discussion:

Both internships discussed in detail
Quantization, LoRA, QLoRA, fine-tuning
OCR and PII detection models — internal workings and comparisons
Annotation tools/models: Roboflow, YOLO vs. RF-DETR comparison
Neural network fundamentals: CNNs, RNNs

Vector Concepts:

Vector Matching: Cosine Similarity, Dot Product, Euclidean Distance
Vector Search: Flat (brute force), IVF (clustering), HNSW (graph-based)

🧠 Scenario Question:

"You have 1 million document vectors. How would you efficiently find the Top-10 most similar vectors for a query?"

HR Discussion (same round):

Achievements, strengths, weaknesses
Why Uptiq, and general awareness about the company

✅ Result: 8 candidates selected for Internship + Performance-based PPO, and 8 for Internship + Confirmed PPO.

💡 Key Observation

I wasn't asked any DSA, SQL, or System Design questions in either interview round — the focus stayed almost entirely on ML/AI depth (RAG, vector search, fine-tuning, OCR/PII) matching my R&D profile. Candidates interviewing for SDE roles were likely tested more heavily on DSA and system design instead.