diff --git a/README.md b/README.md
index 44e0723..2e9d432 100644
--- a/README.md
+++ b/README.md
@@ -91,3 +91,5 @@ scripts/
 ```
 
 Each round's `data/` directory is generated locally and gitignored.
+This is <your-name>'s PR
+This is Bogomil's PR
diff --git a/rounds/3_dna/solution.py b/rounds/3_dna/solution.py
index 8b917da..3970735 100644
--- a/rounds/3_dna/solution.py
+++ b/rounds/3_dna/solution.py
@@ -5,13 +5,119 @@
 own faster implementation.
 """
 
-from .baseline import find_matches as _baseline
+from __future__ import annotations
 
+from concurrent.futures import ThreadPoolExecutor
+from typing import Iterable
 
-def find_matches(fasta_path: str, pattern: bytes) -> list[tuple[str, list[int]]]:
-    """Find every FASTA record whose sequence contains ``pattern``.
 
-    Returns ``[(record_id, [positions...]), ...]`` in file order.
+def _search_record(
+    item: tuple[int, str, str],
+    pattern: str,
+) -> tuple[int, str, list[int]] | None:
     """
-    # TODO: remove this delegation and write your own implementation here.
-    return _baseline(fasta_path, pattern)
+    Search one FASTA record.
+
+    Returns:
+        (original_index, record_id, positions)
+        or None if no matches.
+    """
+    index, record_id, sequence = item
+
+    positions: list[int] = []
+    start = 0
+
+    while True:
+        pos = sequence.find(pattern, start)
+        if pos == -1:
+            break
+
+        positions.append(pos)
+
+        # advance by 1 so overlapping matches count
+        start = pos + 1
+
+    if positions:
+        return (index, record_id, positions)
+
+    return None
+
+
+def _parse_fasta(path: str) -> Iterable[tuple[int, str, str]]:
+    """
+    Stream FASTA records one at a time.
+
+    Yields:
+        (record_index, record_id, sequence)
+    """
+    with open(path, "r") as f:
+        record_id = None
+        seq_parts: list[str] = []
+        index = 0
+
+        for line in f:
+            line = line.strip()
+
+            if not line:
+                continue
+
+            if line.startswith(">"):
+                # emit previous record
+                if record_id is not None:
+                    yield (
+                        index,
+                        record_id,
+                        "".join(seq_parts),
+                    )
+                    index += 1
+
+                record_id = line[1:].strip()
+                seq_parts = []
+
+            else:
+                seq_parts.append(line)
+
+        # emit final record
+        if record_id is not None:
+            yield (
+                index,
+                record_id,
+                "".join(seq_parts),
+            )
+
+
+def find_matches(
+    fasta_path: str,
+    pattern: bytes,
+) -> list[tuple[str, list[int]]]:
+    """
+    Find every FASTA record whose sequence contains pattern.
+
+    Returns:
+        [(record_id, [positions...]), ...]
+    """
+    pattern_str = pattern.decode("ascii")
+
+    results: list[tuple[int, str, list[int]]] = []
+
+    # free-threaded Python can actually parallelize this
+    with ThreadPoolExecutor() as pool:
+
+        futures = [
+            pool.submit(_search_record, record, pattern_str)
+            for record in _parse_fasta(fasta_path)
+        ]
+
+        for future in futures:
+            result = future.result()
+
+            if result is not None:
+                results.append(result)
+
+    # preserve FASTA file order
+    results.sort(key=lambda x: x[0])
+
+    return [
+        (record_id, positions)
+        for _, record_id, positions in results
+    ]
\ No newline at end of file