🔬 Applied Scientist & ML Specialist938 Problems · 127 Topics🌐 Linked to Tech Vault

TalentMe Science Depth: Universal Applied Science & ML Depth Problem Bank

938 end-to-end science depth, system design, transition bridge & BQ problems covering derivations, production trade-offs, pitfalls, and follow-ups.

🎯

Science Depth Mastery Tracker

0% Mastered

Locally synced in your browser. Track, review, and master all 938 scientific & design topics.

Mastered: 0
Reviewing: 0
Unstarted: 102
🚀 Start Next Unpracticed →
🌐
Linked to TalentMe Tech Vault & MindmapsM7: Retrieval, Ranking & RecSys
All 938 questions are directly linked to TalentMe's Tech Vault guides, atomic cards, and interactive mindmaps.
🧭RecSys & Search System Design Mindmap→📖Industrial RecSys & Multi-Task Ranking→🎯5-Step System Design→
Found 24 in-depth questions
M7-001Retrieval, Ranking & RecSysSparse Retrieval (BM25 / TF-IDF)
Easy

Write the BM25 Formula and Explain the Roles of Hyperparameters k1 and b

⚡ One-Sentence Answer

BM25 weights terms using saturated term frequency with document length normalization and inverted document frequency (IDF), where k1 modulates the term frequency saturation curve and b governs the strength of document length penalty.

Math & Tradeoffs→
M7-002Retrieval, Ranking & RecSysSparse Retrieval (BM25 / TF-IDF)
Easy

Explain the Structure and Query Processing Pipeline of an Inverted Index

⚡ One-Sentence Answer

An inverted index maps vocabulary terms to posting lists containing document IDs, term frequencies, and positions; query processing intersects or unions posting lists before ranking documents via scoring algorithms.

Math & Tradeoffs→
M7-003Retrieval, Ranking & RecSysSparse Retrieval (BM25 / TF-IDF)
Medium

Explain the Motivation Behind TF-IDF and Its Two Constituent Components

⚡ One-Sentence Answer

TF evaluates a term's local relevance within a specific document, while IDF measures its global discriminative power across the entire corpus; their product assigns maximum weight to terms that are frequent locally yet rare globally.

Math & Tradeoffs→
M7-004Retrieval, Ranking & RecSysSparse Retrieval (BM25 / TF-IDF)
Medium

Explain the Principles and Architecture of Learned Sparse Retrieval Such as SPLADE

⚡ One-Sentence Answer

SPLADE projects tokens through a BERT masked language modeling (MLM) head with ReLU activations and max-pooling to produce sparse term weights over the vocabulary, combining deep semantic expansion with inverted index execution efficiency.

Math & Tradeoffs→
M7-005Retrieval, Ranking & RecSysSparse Retrieval (BM25 / TF-IDF)
Hard

Explain the Limitations of Sparse Retrieval and Key Directions for Improvement

⚡ One-Sentence Answer

Sparse retrieval is fundamentally constrained by vocabulary mismatch, word order insensitivity, and lack of semantic generalization; key remediation strategies include learned query expansion, learned sparse models, and dense-sparse hybrid retrieval.

Math & Tradeoffs→
M7-006Retrieval, Ranking & RecSysSparse Retrieval (BM25 / TF-IDF)
Hard

Explain the Impact of Document Length on Retrieval Scoring and Common Mitigation Strategies

⚡ One-Sentence Answer

Long documents naturally accumulate higher raw term frequencies and accidental matches; length normalization (such as BM25 parameter b), document chunking, and field weighting eliminate unfair verbosity advantages.

Math & Tradeoffs→
M7-007Retrieval, Ranking & RecSysSparse Retrieval (BM25 / TF-IDF)
Hard

Explain Key BM25 Extensions and Enhancements: BM25+, BM25F, and Learning-to-Rank Weights

⚡ One-Sentence Answer

BM25+ establishes a lower bound on term frequency scores to prevent over-penalizing verbose documents, BM25F computes non-linear saturation over weighted multi-field combinations, and Learning-to-Rank dynamically learns feature weights.

Math & Tradeoffs→
M7-008Retrieval, Ranking & RecSysDense Retrieval & Dual-Encoders
Easy

Explain the Architecture and Training Protocol of Dual Encoder (Bi-Encoder) Retrieval

⚡ One-Sentence Answer

Dual encoder models map queries and documents into a shared vector embedding space via separate or shared encoders, enabling document vectors to be precomputed offline and queried online in sub-millisecond time via approximate nearest neighbor (ANN) search.

Math & Tradeoffs→
M7-009Retrieval, Ranking & RecSysDense Retrieval & Dual-Encoders
Easy

Explain Dense Passage Retrieval (DPR) Training with In-Batch Negatives

⚡ One-Sentence Answer

DPR trains independent query and passage encoders using in-batch negatives (treating other queries' positive passages within the batch as negatives) augmented with BM25 hard negatives to optimize retrieval discriminativeness.

Math & Tradeoffs→
M7-010Retrieval, Ranking & RecSysDense Retrieval & Dual-Encoders
Medium

Explain the Role and Methodology of Hard Negative Mining in Dense Retrieval

⚡ One-Sentence Answer

Random negatives provide negligible gradient signals because the model easily distinguishes them; hard negative mining identifies semantically or lexically similar non-relevant passages to enforce sharp decision boundaries.

Math & Tradeoffs→
M7-011Retrieval, Ranking & RecSysDense Retrieval & Dual-Encoders
Medium

Explain the Temperature Parameter in Contrastive Learning and Its Impact on Optimization

⚡ One-Sentence Answer

The temperature parameter tau scales the logit distribution in the softmax denominator; a smaller tau sharpens the distribution to focus gradients on the hardest negatives, while a larger tau distributes gradients more uniformly across all samples.

Math & Tradeoffs→
M7-012Retrieval, Ranking & RecSysDense Retrieval & Dual-Encoders
Hard

Explain the Trade-Off Between Embedding Dimensionality and Storage/Compute Costs in Dense Retrieval

⚡ One-Sentence Answer

Higher embedding dimensions improve semantic capacity but scale index storage and ANN latency linearly; techniques like Matryoshka Representation Learning (MRL), Product Quantization (PQ), and binary quantization optimize the Pareto frontier.

Math & Tradeoffs→
M7-013Retrieval, Ranking & RecSysDense Retrieval & Dual-Encoders
Hard

Explain the Key Challenges and Solutions in Cross-Modal Dense Retrieval

⚡ One-Sentence Answer

Cross-modal retrieval projects disparate modalities (text, images, audio) into a shared embedding space, facing core hurdles including modality gaps, noisy training alignment, and preserving fine-grained details against global semantics.

Math & Tradeoffs→
M7-014Retrieval, Ranking & RecSysDense Retrieval & Dual-Encoders
Hard

Explain Instruction-Based and Task-Specific Dense Embeddings in Modern Retrievers

⚡ One-Sentence Answer

Instruction-based embedding models (e.g., E5, BGE, Instructor) prepend task instructions or prefixes to inputs, enabling a single unified encoder to dynamically optimize representations for disparate tasks like retrieval, clustering, and classification.

Math & Tradeoffs→
M7-015Retrieval, Ranking & RecSysDense Retrieval & Dual-Encoders
Hard

Explain the Problem of False Negatives in Dense Retrieval and Debiasing Strategies

⚡ One-Sentence Answer

In-batch and mined negatives frequently contain unlabeled yet genuinely relevant passages; treating them as negative samples injects destructive gradient penalties that suppress relevant content, necessitating algorithmic debiasing and filtering.

Math & Tradeoffs→
M7-016Retrieval, Ranking & RecSysDense Retrieval & Dual-Encoders
Hard

Explain Domain Adaptation and Fine-Tuning Strategies for Dense Retrieval Models

⚡ One-Sentence Answer

General-domain embeddings degrade under specialized domain vocabularies (medical, legal, code); unsupervised masked language model adaptation, synthetic query generation (GPL), and contrastive fine-tuning with hard negatives restore state-of-the-art recall.

Math & Tradeoffs→
M7-017Retrieval, Ranking & RecSysDense Retrieval & Dual-Encoders
Hard

Explain Multilingual and Cross-Lingual Dense Retrieval Architectures and Alignment Techniques

⚡ One-Sentence Answer

Multilingual dense retrieval maps queries and documents from diverse languages into a shared vector space, enabling direct cross-lingual matching (e.g., Chinese query retrieving English documents) without runtime machine translation.

Math & Tradeoffs→
M7-018Retrieval, Ranking & RecSysHybrid Retrieval & RRF Fusion
Easy

Explain the Principles of Reciprocal Rank Fusion (RRF) and the Role of Hyperparameter k

⚡ One-Sentence Answer

RRF fuses multiple retrieval candidate lists by summing the reciprocal of each item's rank offset by constant k, providing scale-invariant, calibration-free rank aggregation that is highly robust to score distribution disparities.

Math & Tradeoffs→
M7-019Retrieval, Ranking & RecSysHybrid Retrieval & RRF Fusion
Easy

Explain Why Hybrid Retrieval (Sparse + Dense) Consistently Outperforms Single Retrieval Modalities

⚡ One-Sentence Answer

Hybrid retrieval unites the exact lexical matching, zero-shot robustness, and identifier precision of sparse search with the semantic abstraction, synonym generalization, and multi-hop reasoning of dense embeddings, eliminating blind spots across heterogeneous query distributions.

Math & Tradeoffs→
M7-020Retrieval, Ranking & RecSysHybrid Retrieval & RRF Fusion
Medium

Explain the Differences Between Score-Based Fusion and Rank-Based Fusion in Search Systems

⚡ One-Sentence Answer

Score-based fusion combines normalized numerical scores to preserve model confidence margins but requires rigorous calibration; rank-based fusion (such as RRF) aggregates ordinal rankings, offering complete scale invariance and robustness at the expense of discarding margin magnitudes.

Math & Tradeoffs→
M7-021Retrieval, Ranking & RecSysHybrid Retrieval & RRF Fusion
Medium

Explain Multi-Channel Candidate Generation (Multi-Channel Retrieval) and Quota Allocation Strategies

⚡ One-Sentence Answer

Multi-channel retrieval combines heterogeneous candidate sources (lexical, semantic, collaborative filtering, popular, real-time) to maximize candidate diversity, distributing candidate quotas based on incremental marginal recall under downstream re-ranking latency constraints.

Math & Tradeoffs→
M7-022Retrieval, Ranking & RecSysHybrid Retrieval & RRF Fusion
Hard

Explain Query Rewriting and Expansion Techniques in Modern Information Retrieval

⚡ One-Sentence Answer

Query rewriting refines, disambiguates, and decomposes user queries, while query expansion appends synonyms and related terms; techniques range from classical pseudo-relevance feedback (PRF) to LLM-generated hypothetical documents (HyDE).

Math & Tradeoffs→
M7-023Retrieval, Ranking & RecSysHybrid Retrieval & RRF Fusion
Hard

Explain Training-Serving Consistency and Feature Skew in Search and Retrieval Systems

⚡ One-Sentence Answer

Training-serving skew occurs when offline training data diverges from real-time online inference conditions across features, text preprocessing, or model versions, serving as one of the most pervasive root causes of production search degradation.

Math & Tradeoffs→
M7-024Retrieval, Ranking & RecSysHybrid Retrieval & RRF Fusion
Hard

Explain Learned Fusion and Weight Optimization in Multi-Channel Retrieval Systems

⚡ One-Sentence Answer

Learned fusion replaces heuristic weights and static rank formulas with a machine-learned ranking model (e.g., LambdaMART or shallow neural rankers) that dynamically combines multi-channel scores and contextual features to maximize ranking metrics.

Math & Tradeoffs→