Executive Summary

Andy Zaitsev presents as a technically strong ML engineer whose core competencies — production inference pipelines, RAG systems, model efficiency, and published research — are genuinely relevant to Pergola Studio's mission. Their efficiency-first mindset (MossNet, low-latency Rust systems) mirrors the company's cost-reduction thesis. However, the role's explicit requirements around agentic frameworks (LangGraph, LangSmith, CrewAI, MCP) represent a meaningful gap that is not addressed anywhere in their resume. More critically, the LinkedIn URL submitted resolves to Sergei Sergienko, CEO of Pivots Global (the hiring organization), which is an unexplained and material inconsistency that must be clarified before any further evaluation. If the identity issue is resolved and the candidate can demonstrate even foundational agentic framework exposure, they could be a strong BORDERLINE-to-FIT candidate worth a technical interview.

Top Strengths

✓Production-grade ML engineering with measurable, quantified results at scale
✓Published ML research on arXiv — directly satisfies a key differentiator the job listing calls out
✓Efficiency-focused model design (MossNet) aligns directly with Pergola Studio's cost-reduction mission
✓Full MLOps lifecycle experience: training, CI/CD, drift detection, deployment, monitoring
✓Polyglot systems engineer (Python + Rust) capable of owning both ML and infrastructure layers

Key Concerns

!LinkedIn profile mismatch is a serious integrity flag — the submitted URL belongs to the hiring company's CEO, not the applicant
!No demonstrated experience with explicitly required agentic frameworks (LangGraph, LangSmith, CrewAI, MCP servers) — the largest technical gap for this role

Culture Fit

55%

Growth Potential

High

Salary Estimate

$90,000 - $120,000 (aligns with lower-to-mid band given 5 years experience and EU-based location)

Assessment Reasoning

BORDERLINE decision is based on two competing signals. On the positive side, Andy Zaitsev's resume demonstrates genuine senior-level ML engineering capability, production AI system ownership, efficiency-focused model design, and published research — all of which directly map to core aspects of this Founding Engineer role. On the negative side, the most explicitly listed required skills (LangGraph, LangSmith, LangFuse, CrewAI, MCP servers, agent orchestration) are entirely absent from their profile, representing a ~40% gap in the technical skill checklist. The LinkedIn URL mismatch (resolving to the hiring company's own CEO) introduces an integrity concern that alone could warrant rejection pending explanation. The candidate is not a clear NOT_FIT due to strong underlying ML engineering fundamentals and research credibility, but cannot be rated FIT without resolving the identity discrepancy and gaining clarity on agentic framework experience. HR review and an initial screening call to resolve the LinkedIn issue are recommended before committing to a technical interview.

Interview Focus Areas

Clarify LinkedIn URL discrepancy — confirm identity and correct profileDeep-dive on agentic framework experience: has the candidate worked with LangGraph, LangSmith, or similar? Any plans or adjacent experience?Assess ability to transition from DeFi/trading domain to marketing AIExplore MossNet in detail — training methodology, distillation techniques, evaluation approachEvaluate founding-team readiness: comfort with ambiguity, strategic input, and leadership in a 0-to-1 environment

Code Review

FairSenior Level

No code sample was provided in the application, limiting direct code quality assessment. Based on project descriptions — especially the IRIS liquidation engine in Rust and MossNet edge model — the candidate appears to operate at a senior engineering level with attention to performance and architecture. A technical screen reviewing the referenced GitHub repository is strongly recommended before drawing conclusions.

PythonRustPyTorchFastAPITerraformClickHouseFAISS

+GitHub profile referenced (github.com/0xf104a) and personal site (f104a.io) suggest publicly visible work
+Project descriptions (MossNet, GlacierX, IRIS/Darlingtonia) demonstrate architectural thinking and production discipline

-No code example was submitted with the application — direct code quality cannot be assessed
-GitHub was not linked in the application form, so repository depth/activity is unverifiable at this stage

Experience Overview

5y total · 4y relevant

Andy Zaitsev is a capable ML Engineer with 5 years of production AI/ML experience, strong Python and Rust skills, and a published arXiv paper. Their infrastructure depth, low-latency ML pipeline work, and edge model efficiency research align well with Pergola Studio's cost-efficiency mission. However, their resume shows no hands-on experience with the agentic frameworks (LangGraph, LangSmith, CrewAI) or MCP/tool-calling patterns that are core requirements for this role.

Matching Skills

Python (advanced)LLM integrationRAG architecturesSemantic search / embeddingsPrompt engineeringHuggingFace TransformersPyTorch / fine-tuningNumPy / PandasFAISS vector searchLlamaIndex (familiar)FastAPI / microservicesAWS (EC2, S3, Lambda)Docker / Kubernetes / TerraformMLflow / MLOps / CI-CDDistributed systems & low-latency inferencePublished ML research (arXiv)

Skills to Verify

LangGraph / LangSmith / LangFuse (not mentioned)CrewAI (not mentioned)Agentic frameworks / agent orchestration (no direct evidence)MCP servers and tool calling (not mentioned)Multimodal model integration (not demonstrated)AI observability tooling (e.g., LangFuse, Helicone)Marketing-domain AI experienceModel distillation / compression at production scale

Candidate information is anonymized. Personal details are hidden for fair evaluation.