1

NLP & Language Models

Practical NLP work at the level required by research projects — task-specific fine-tuning, embedding-based retrieval, RAG systems, and instruction-tuned model pipelines.

Sub-topics and representative methods
Understanding & IE BERT-family encoders, span-extraction pipelines, NER / relation extraction, structured-output decoding
Generation & instruction tuning Llama-family / Qwen-family / Mistral-family base and instruct models, SFT, LoRA / QLoRA, DPO
Embeddings & retrieval BGE, E5, Instructor-style embeddings, ColBERT and late-interaction retrieval, hybrid BM25 + dense
RAG systems Chunking strategies, multi-vector retrieval, reranking, evaluation harnesses (Ragas-style), tool-augmented generation
2

Machine Learning & Deep Learning

General ML/DL work that spans multiple domains — representation learning, self-supervised methods, reinforcement learning, and the underlying training-pipeline engineering.

Sub-topics and representative methods
Supervised learning Standard tabular and image classification pipelines, calibration, robust training
Self-supervised & representation Contrastive methods (SimCLR, MoCo), masked-image modeling (MAE), masked-language modeling, multi-modal contrastive
Reinforcement learning PPO, SAC, offline RL (CQL-style), RLHF-style preference pipelines
Training craft Mixed-precision training, gradient accumulation, deterministic training, distributed data parallel
3

Graph Algorithms & Recommendation

Graph-structured methods, both classical algorithmic and neural, as well as recommendation systems where graph and embedding methods intersect.

Sub-topics and representative methods
Graph neural networks GCN, GAT, GraphSAGE, message-passing frameworks, heterogeneous-graph models
Classical graph algorithms Shortest paths, community detection, graph matching, max-flow / min-cut
Recommendation systems Two-tower retrieval, SASRec / BERT4Rec sequential models, PinSAGE-style large-graph retrieval, learning-to-rank

If your work is adjacent to one of the areas above — a different sub-topic, a different application domain, an overlap between two — please contact us. We will review the materials you send and respond with one of three answers: acceptance, decline, or "additional literature review is required before a decision."

To discuss a paper, draft, or prototype: ✉ bd@sciwealth.com