<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="mikhailov-etal-2026-multi">
    <titleInfo>
        <title>Multi-Specialist LLM Relay System for Competitive Programming</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Andrei</namePart>
        <namePart type="family">Mikhailov</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Mikhail</namePart>
        <namePart type="family">Burtsev</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Alsu</namePart>
        <namePart type="family">Sagirova</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-10</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Hung-Yi</namePart>
            <namePart type="family">Lee</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Sunipa</namePart>
            <namePart type="family">Dev</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jessy</namePart>
            <namePart type="family">Li</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Desmond</namePart>
            <namePart type="family">Elliott</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Budapest, Hungary</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-459-0</identifier>
    </relatedItem>
    <abstract>Large Language Models excel at code generation, yet competitive programming exposes a persistent failure mode: existing multi-agent pipelines distribute work over generic planner, coder, and debugger roles and delegate the choice of algorithmic technique to the backbone alone. We present MARS (Multi-Agent Relay of Specialized LLMs), a prompt-only framework in which each agent is a topic specialist—dynamic programming, graphs, strings, geometry, and so on—grounded by retrieval-augmented generation over an algorithm-theory corpus. Given a problem, retrieval selects a small team of relevant specialists; a starter writes an initial C++17 solution, and each subsequent turn runs the candidate against public examples in a sandbox, lets the active specialist keep, repair, or hand off the draft, and forwards a structured packet to the next specialist. A single infrastructure-fixer pass normalizes boilerplate at the end. On the CodeContests test split with Gemma 4, MARS reaches 0.624 \pm 0.006 pass rate at 2.3 relay steps per task, +13.9pp over direct prompting, closing most of the gap to CodeSIM at 0.731 with 3.3\times lower wall-clock cost and substantially smaller variance in per-task token spend.</abstract>
    <identifier type="citekey">mikhailov-etal-2026-multi</identifier>
    <identifier type="doi">10.18653/2026.emnlp-main.1359</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.1359/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>30995</start>
            <end>31007</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Multi-Specialist LLM Relay System for Competitive Programming
%A Mikhailov, Andrei
%A Burtsev, Mikhail
%A Sagirova, Alsu
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F mikhailov-etal-2026-multi
%X Large Language Models excel at code generation, yet competitive programming exposes a persistent failure mode: existing multi-agent pipelines distribute work over generic planner, coder, and debugger roles and delegate the choice of algorithmic technique to the backbone alone. We present MARS (Multi-Agent Relay of Specialized LLMs), a prompt-only framework in which each agent is a topic specialist—dynamic programming, graphs, strings, geometry, and so on—grounded by retrieval-augmented generation over an algorithm-theory corpus. Given a problem, retrieval selects a small team of relevant specialists; a starter writes an initial C++17 solution, and each subsequent turn runs the candidate against public examples in a sandbox, lets the active specialist keep, repair, or hand off the draft, and forwards a structured packet to the next specialist. A single infrastructure-fixer pass normalizes boilerplate at the end. On the CodeContests test split with Gemma 4, MARS reaches 0.624 \pm 0.006 pass rate at 2.3 relay steps per task, +13.9pp over direct prompting, closing most of the gap to CodeSIM at 0.731 with 3.3\times lower wall-clock cost and substantially smaller variance in per-task token spend.
%R 10.18653/2026.emnlp-main.1359
%U https://aclanthology.org/2026.emnlp-main.1359/
%U https://doi.org/10.18653/2026.emnlp-main.1359
%P 30995-31007

Markdown (Informal)

[Multi-Specialist LLM Relay System for Competitive Programming](https://aclanthology.org/2026.emnlp-main.1359/) (Mikhailov et al., EMNLP 2026)

• Multi-Specialist LLM Relay System for Competitive Programming (Mikhailov et al., EMNLP 2026)

ACL

• Andrei Mikhailov, Mikhail Burtsev, and Alsu Sagirova. 2026. Multi-Specialist LLM Relay System for Competitive Programming. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 30995–31007, Budapest, Hungary. Association for Computational Linguistics.